low-power shared access to spectrum for mobile …€¦ · low-power shared access to spectrum for...

Real Wireless Limited. Company Registered in England&Wales no. 6016945. Registered Office: 94 New Bond Street, London, W1S 1SJ

Final Report

Low-power shared access to spectrum for mobile

broadband

Ofcom Project MC/073

Issued to: Draft issued to Ofcom for review Version: 1.0 Issue Date: 24th January 2011

Real Wireless Ltd.

PO Box 2218

Pulborough

West Sussex

RH20 4XB

United Kingdom

www.realwireless.biz

[email protected]

Tel: +44 207 117 8514

Fax: +44 808 280 0142

http://www.realwireless.biz/

mailto:[email protected]

2 Final Report: Low Power Shared Access to Mobile Broadband Spectrum

About Real Wireless

Real Wireless is a leading independent wireless consultancy, based in the U.K. and

working internationally for enterprises, vendors, operators and regulators – indeed

any organization which is serious about getting the best from wireless to the benefit

of their business. We seek to demystify wireless and help our customers get the

best from it, by understanding their business needs and using our deep knowledge

of wireless to create an effective wireless strategy, implementation plan and

management process. We are experts in radio propagation, international spectrum

regulation, wireless infrastructures, and much more besides. We have experience

working at senior levels in vendors, operators, regulators and academia.

We have specific experience in LTE, UMTS, HSPA, Wi-Fi, WiMAX, DAB, DTT,

GSM, TETRA – and many more.

For details contact us at: [email protected]

Tap into our news and views at: realwireless.wordpress.com

Stay in touch via our tweets at twitter.com/real_wireless

Keep up to date with 4G news at realwireless4g.wordpress.com



http://realwireless.wordpress.com/

http://twitter.com/real_wireless

http://twitter.com/real_wireless

http://realwireless4g.wordpress.com/

http://realwireless4g.wordpress.com/


Contents Final Report ......................................................................................................................................................... 1

Low-power shared access to spectrum for mobile broadband....................................................................... 1

1 Executive Summary ..................................................................................................................................... 8

2 Introduction ................................................................................................................................................. 9

2.1 UK spectrum at 2.6GHz is due to be auctioned shortly....................................................................... 9

2.2 Auctioning of the 2.6GHz band is complicated by small cells, interference to adjacent bands and

maintaining competition ............................................................................................................................... 10

2.3 Ofcom want to understand the technical issues associated with low-power shared access in 2.6GHz

spectrum ........................................................................................................................................................ 11

2.4 Our approach and assumptions ........................................................................................................ 11

2.5 The evolution of small cell technologies and the opportunity for more efficient spectrum usage .. 13

3 Our results indicate that a low power shared access channel is feasible if standard interference

mitigation techniques are applied ..................................................................................................................... 15

3.1 A maximum transmit power of 30 dBm is recommended in a 2 x 20MHx scenario ......................... 16

3.2 Interference amongst low power operators is manageable given cooperation on mitigation

techniques and conventions and a height limit on outdoor deployments ................................................... 16

3.3 An underlay approach presents interference and coordination challenges, so we recommend a

hybrid approach if dedicated 2x20 MHz spectrum is not feasible ................................................................ 17

3.4 Our recommendations rely on the use of interference mitigation techniques and other

coordination measures, facilitated by a code of practice agreed amongst licensees .................................. 18

3.5 We recommend positioning the shared channel at the upper end of the FDD block, although

further study is needed ................................................................................................................................. 19

3.6 Lessons learnt from the DECT guard band study and Wi-Fi .............................................................. 19

3.6.1 The DECT Guard Band ................................................................................................................ 19

3.6.2 Wi-Fi in the 2.4 GHz band .......................................................................................................... 21

4 Our assumptions on LTE devices and interference mitigation techniques ............................................... 24

4.1 LTE device parameters ...................................................................................................................... 24

4.2 Quality of service and interference criteria assumptions ................................................................. 28


4.3 Typical building sizes in target deployments ..................................................................................... 28

4.4 Propagation models used .................................................................................................................. 28

4.4.1 Path types to be modelled ........................................................................................................ 29

4.4.2 Models used in other studies .................................................................................................... 29

4.4.3 Selection of models according to path type .............................................................................. 31

4.5 Small cell interference mitigation techniques ................................................................................... 32

4.5.1 Control Channel Interference in LTE ......................................................................................... 34

4.5.2 Femto Forum study of interference management for OFDMA femtocells ............................... 35

4.5.3 3GPP Technical report ............................................................................................................... 38

4.6 Stakeholder views .............................................................................................................................. 45

4.7 We have made reasonable simplifying assumptions regarding interference mitigation techniques

based on the literature and stakeholder feedback ....................................................................................... 47

5 Power levels in line with 3GPP recommendations will ensure good coverage for low-power shared

access points ...................................................................................................................................................... 50

5.1 Study questions ................................................................................................................................. 50

5.2 Indoor office coverage – Study question 1.13 ................................................................................... 52

5.2.1 Scenario description and assumptions ...................................................................................... 52

5.2.2 Results and conclusions ............................................................................................................. 53

5.3 Residential coverage – Study question 1.13 ..................................................................................... 58



5.4 Indoor public area coverage – Study question 1.13 .......................................................................... 71



5.5 Business park/Campus environment – Study question 1.13 ............................................................. 76



5.6 Depth of in-building coverage – Study question 1.14 ....................................................................... 80


5.6.1 Approach to depth of in-building coverage ............................................................................... 80


5.7 Summary ............................................................................................................................................ 87

6 Co channel interference is manageable via appropriate antenna heights, dynamic scheduling and

compromises in target throughputs .................................................................................................................. 89

6.1 Study questions ................................................................................................................................. 89

6.2 Building separation – Study question 1.17 ........................................................................................ 92



6.3 Outdoor antenna height – study question 1.18 ................................................................................ 96



6.4 Interference to macrocells in an underlay approach – study question 1.19 .................................... 98


6.4.2 Results and conclusions ........................................................................................................... 100

6.5 Separation distance between outdoor and indoor low power networks – study question 1.20 ... 103

6.5.1 Scenario description and assumptions .................................................................................... 103

6.5.2 Results and conclusions ........................................................................................................... 104

6.6 Summary .......................................................................................................................................... 107

7 Trade-offs relating to spectrum quantity ................................................................................................ 110

7.1 Study questions ............................................................................................................................... 110

7.2 Factors impacting capacity in small cells ......................................................................................... 112

7.2.1 Spectrum efficiency density in small cells ............................................................................... 112

7.2.2 The impact of uncoordinated shared access on capacity ....................................................... 116

7.2.3 Summary of trade offs across spectrum options .................................................................... 118

7.3 Example scenarios of how capacity is impacted by spectrum choice ............................................. 120

7.3.1 Residential scenario ................................................................................................................. 121

7.3.2 Shopping centre scenario ........................................................................................................ 122


7.4 Summary .......................................................................................................................................... 123

8 Adjacent channel interference from TDD and S-band radar is less significant than from FDD macros.. 126

8.1 Study questions ............................................................................................................................... 126

8.2 TDD interference ............................................................................................................................. 129

8.2.1 Scenario 18 – Impact of interference from adjacent WiMAX or TD-LTE network .................. 129

8.2.2 Results from analysis of TDD ACI into FDD low power access point ....................................... 131

8.2.3 Review of previous studies and conclusions ........................................................................... 133

8.3 S-band radar interference ............................................................................................................... 136

8.3.1 Scenario 19 – Impact of interference from radar emission in the 2.7-2.9 GHz band to

operation of low power networks ........................................................................................................... 136

8.3.2 Results from separation distance between an airport (ATC) radar and FDD UE..................... 137

8.3.3 Review of previous studies and conclusions ........................................................................... 140

8.4 Summary .......................................................................................................................................... 142

9 Conclusions and recommendations ........................................................................................................ 144

9.1 Overall ............................................................................................................................................. 144

9.2 Coverage .......................................................................................................................................... 145

9.3 Co-channel Interference .................................................................................................................. 146

9.4 Adjacent Channel Interference ....................................................................................................... 147

9.5 Trade-offs relating to spectrum quantity ........................................................................................ 148

9.5.1 10MHz Vs. 20MHz ................................................................................................................... 148

9.5.2 Number of operators ............................................................................................................... 149

9.5.3 Dedicated vs. underlay vs. hybrid ............................................................................................ 149

10 Appendix A – Detailed description of propagation models ................................................................ 153

10.1 ITU-R P. 1411 ................................................................................................................................... 153

10.2 COST 231 Walfisch-Ikegami ............................................................................................................. 153

10.3 COST 231 LoS Building Penetration Model...................................................................................... 153

10.4 COST 231 Multi-Wall Model ............................................................................................................ 153

10.5 Modelling for shadowing (slow fading) ........................................................................................... 153


11 References ........................................................................................................................................... 154


1 Executive Summary


2 Introduction

2.1 UK spectrum at 2.6GHz is due to be auctioned shortly

Ofcom is seeking to auction licenses for the frequency band 2500 MHz – 2690 MHz (the “2.6 GHz”

band), with awards of licenses expected to take place in 2012. The band is arranged as 2500 – 2570

MHz (uplink) paired with 2620 – 2690 MHz (downlink), plus an unpaired portion 2570 – 2620 MHz.

Although the award will be on a technology neutral basis, the expected technologies are LTE in

frequency division duplex mode in the paired spectrum and WiMAX or TD-LTE in the unpaired

spectrum.

The 2.6GHz band is recognised as a highly desirable band for cellular operators and vendors, as its

usage has been commonly defined for International Mobile Telecommunications by the

International Telecommunications Union in all three of its regions [1]. This international

harmonisation of the band allows vendors and operators to benefit from greater economies of scale.

Sometimes known as the UMTS expansion band, it was originally anticipated that this band would be

used by cellular operators to provide additional capacity to 3G networks [1]. However, more recently

it has been associated with 4G networks and indeed is already being used in Sweden, Norway,

Uzbekistan and Hong Kong to provide LTE services and in several places including the US to provide

WiMAX services [2].

In Europe, Commission Decision 2008/477/EC outlined partitioning of this band between TDD and

FDD services and Block Edge Mask (BEM) parameters for devices within this band. Issued in June

2008, this decision recommends that the 2.6GHz band should be made available by all member

states within six months. This band has already been auctioned in many European countries

including Denmark, Finland, Germany, Norway, the Netherlands and Sweden [1]. As an EU member

the UK must also act upon this decision.

Ofcom has outlined an ambitious timetable for auctioning of the 2.6GHz band which includes

publishing a consultation on this subject by the end of February 2011 which will end by May [3]. It is

anticipated that the final auction regulations will be in place by the end of 2011 with bidding

occurring in 2012 and spectrum becoming available for first deployments by 2013.

This study is one element of analysis as input to that consultation.It relates to the potential to

designate part of the spectrum for shared low-power use. In contrast to the remainder of the

spectrum, this low-power shared access would involve multiple licensees operating in a concurrent,

non-exclusive fashion across the same spectrum. The spectrum could be awarded on an exclusive


basis, where only low-power licensees would have access, or potentially on an underlay basis, where

low-power shared access uses the same spectrum as a wide area (high-power) licence.

There is some precedent for such a licence in Ofcom’s previous award of low-power concurrent

licences in the 1781.1-1785 MHz and 1876.7 MHz – 1880 bands, (often known as the DECT guard

band4). The auction outcome was that 12 operators gained concurrent access to 2 x 3.3 MHz of

spectrum. Although technology neutral, most use of this band to date has been for small area low-

power GSM systems.

2.2 Auctioning of the 2.6GHz band is complicated by small cells, interference

to adjacent bands and maintainingcompetition

The low-power licenses exhibit particular potential interference challengesamongst licensees and

from adjacent spectrum users for a number of reasons:

Low-power applications may include femtocells and picocells, where deployment may be conducted by the end user or other untrained personnel, resulting in deployments of cells which may be suboptimum with regards to coverage or interference

The close proximity (in frequency and location) of high-power cells, which could cause substantial interference to the cells or user devices operating in the low-power spectrum.

The presence of multiple concurrent operators in the same spectrum, who may or may not coordinate deployments amongst themselves.

The presence of TDD systems adjacent(in frequency and location) to FDD systems in paired spectrum, and the presence of high-power radar systems adjacent to FDD systems in paired spectrum.

In order to limit these interference sources, the low-power licences will include appropriate

technical conditions which will limit interference. The licensees might also enter into some form of

agreement to further limit interference which may include, for example, additional technical

restrictions, coordination procedures, shared databases, technical interfaces or roaming

arrangements.

In the DECT guard band situation, the use of 200 kHz GSM channels meant that some frequency

planning amongst operators could avoid most potential cases of interference and issues would only

be experienced at high deployment densities.


2.3 Ofcom want to understand the technical issues associated with low-

power shared access in 2.6GHz spectrum

to investigate the technical issues associated with low-power shared access in the 2.6 GHz spectrum,

which could potentially open up the opportunity for more operators to compete for service

provision to the benefit of citizen-consumers and for the spectrum to be used in a more spectrally

efficient manner and via innovative business models. Ofcom seeks technical investigation to inform

the nature of any such licensing:

Should the low-power access be exclusive to a portion of the band, or on an ‘underlay’ basis, shared with a wide-area high-power licence?

Does the low-power block need 2 x 20 MHz, or is 2 x 10 MHz sufficient? The investigation should account for the traffic capacity, number of concurrent operators, and frequency re-use characteristics of the resulting low-power uses.

What techniques could be used to manage spectrum sharing between low-power operators?

What power level would be needed to provide coverage in a variety of indoor locations, from both indoor antennas and from outdoors?

At what separation distances between buildings does coordination between low-power operators become necessary?

Is 5m outdoor antenna height an appropriate limit for the purpose of limiting interference between low-power networks?

What restrictions would need to be placed on antennas to minimise interference in the case of an underlay network?

What separation is needed between outdoor and indoor low-power antennas to allow frequency reuse and at what distance does operator coordination becomes necessary?

What is the impact of adjacent channel interference from TDD WiMAX or TD-LTE systems and from radar systems to low-power operation and what associated technical conditions may be necessary?

2.4 Our approach and assumptions

Our work in this study was organised into four interdependent workpackages, as illustrated in the

study logic diagram shown in Figure 2-1.


Figure 2-1: Study Logic

We have made some overall assumptions for the purpose of our study:

Although the 2.6 GHz award will be on a technology neutral basis, we have assumed the use of FDD LTE technology in the low power band.

We are not aiming to determine the typical overall performance of systems in the low-power band. Instead we seek to provide analysis relevant to the setting of technical licence conditions, which typically focus on relatively extreme parameter setting and challenging scenarios to set licence conditions and challenging cases. Operators seeking access for this band are advised to conduct system simulations (and field trials where possible) to determine the overall system performance.

We are generally using modelling techniques which are based on identifying the most important paths and static situations. It is not the target here to fully model the whole system performance or the finer details of the interference management techniques which could be applied.

Interference from unpaired spectrum assumes the use of TD-LTE technology in the unpaired band, but this is intended also to be reasonably representative of interference from WiMAX systems.

Building types

Outdoor vs. indoor

Exclusive low-power

vs. underlay

Work package 2 - Define deployment scenarios

Work package 3 - Interference assessment

Interference management

techniques

LTE/WiMAX specifications and

performance

Propagation models

Mathematical models (e.g.

COST231, ITU-R P.1411)

Results review & sensitivity analysis

Parameter recommendations

Reporting

Scenario layouts, equipment parameters & interference criteria

Transmit power, antenna height, & tolerable interference levels

Path losses for coverage

Work package 1 - Capture equipment characteristics

Work package 4

Quantity of spectrum

Path losses for interference paths


2.5 The evolution of small cell technologies and the opportunity for more

efficient spectrum usage

The concept of perhaps five to ten operators sharing the same spectrum for high data rate services,

with cells deployed with limited or no planning, with potentially limited coordination between

operators and possibly sharing with a high power network is undoubtedly ambitious, and a few years

ago may have seemed infeasible. However, in recent years the development of femtocells, originally

intended for deployment in homes, has opened up the possibility for a more flexible use of spectrum

providing high-quality coverage and high capacity in limited areas. Femtocells are low-power access

points, providing services in licenced spectrum which are essentially identical over the air to those

delivered by a conventional base station, and therefore can be delivered to all standard mobile

devices. They were originally designed for the home environment, where they connect into the

mobile operator’s network via the internet, over a standard domestic broadband connection. This

connection allows the operator to remain in management control of the devices, but at the same

time the devices include intelligence permitting, amongst other functions, dynamic management of

interference between cells. Given this dynamic management, operators have increasingly adopted

femtocells, predominantly for the 3G market today – in December 2010 there were 18 commercial

services worldwide and a total of 30 operator commitments to deploy5.

However the same technology is also targeted to application for LTE and WiMAX systems and to

environments beyond the home. Our recent 4G Capacity Gains study for Ofcom6 has highlighted that

low power cells are likely to be a key part of 4G networks. In this study vendors and operators

identified the 2.6GHz band as an excellent candidate for small cell deployments, leaving lower

frequency bands primarily for providing wide area coverage. Similarly, Telefonica have said:

“[Telefonica is moving towards] street-level picocells and femtocells… based on this 2.6-GHz

frequency”

- Jaime Lluch Ladron, TelefonicaNew Technology Executive October 20107

Standards exist for LTE femtocells, known as Home eNodeB (HeNB) in 3GPP specifications, finalised

in 3GPP Release 9 in early 20108. Likewise, WiMAX Forum finalised femtocell specifications in June

20109. Both standards support the 2.6 GHz band.

Due to the low power nature of small cells it may be wasteful to assign separate femtocell carriers to

operators individually. An approach which is coordinated across operators to the minimum extent

necessary to achieve acceptable performance may make a shared low power channel feasible and

improve spectrum utilisation in this band support Ofcom’s duty to ensure the optimal use of the


electro-magnetic spectrum and also to ensure that a wide range of electronic communications

services – including high speed data services – is available throughout the UK10.

Such shared operation, however, relies on the successful operation of relevant interference

management techniques and potentially additionally on operators cooperating to coordinate

parameters and deploymentssuch that each can achieve a viable service offering without excessive

cost or complexity.


3 Our results indicate that a low power shared access channel is

feasible if standard interference mitigation techniques are

applied

A low power shared access channel provides significant potential for making the most of capacity in

the 2.6GHz band due to the improved spectrum efficiency density brought by:

A higher cell density

Improved SINR distribution in small cells

Being typically deployed in indoor environments where MIMO works best

Estimates on the potential capacity improvements with small cells range up to x100.

As small cell access points are low power devices there is a greater opportunity for sharing spectrum

than for traditional wide area cellular networks. This is comparable to the way in which Wi-Fi

systems dynamically share the capacity available, but with a greater degree of flexibility and

assurances of quality of service arising from technical innovations developed for femtocells and the

opportunity for centralised coordination by licenced operators. Also industry results have

highlighted the 2.6GHz band as especially suitable for small cells compared to the low frequency

(sub 1 GHz) bands which are best suited to providing wide area coverage (NEC slides for 4G study

recommending 2.6GHz for small cells).

Our study has found that a low power shared access channel amongst multiple operators is feasible

but will require appropriate restrictions to make best use of the capacity benefits of small cells in the

areas of greatest demand.

We recommend that a 2 x 20MHz channel would be the most attractive solution for a low power

shared access channel at 2.6 GHz. This delivers the maximum user experience, best interference

mitigation amongst operators and allows maximum benefit from capacity gains and spectrum

efficiencies of small cells. If, however, dedicated low power access to this quantity of spectrum

cannot be provided due to the impact on the quantity available to high power licensees, a hybrid

approach where only part of the channel overlaps with a high power allocation will mitigate

potential interference. Restrictions on transmit power, technology and a code of practice relating to

coordination amongst operators are recommended to ensure maximum benefit from this band.


3.1 A maximum transmit power of 30 dBm is recommended in a 2 x 20MHx

scenario

Our results show that, for homes, +20 dBm EIRP gives more than adequate coverage in the majority

of homes, in line with the 3GPP Home e-Node B (LTE femtocells) specification with a low-gain

antenna. In offices and public environments (e.g. shopping centres) somewhat higher power at

around +24 dBm EIRP may be required to provide good coverage at high-end data rates in some

situations. This level is consistent with the HeNB specification with a moderate gain antenna, or the

3GPP Local Area Base station specification

However, in order to provide adequate in-building penetration from outdoor access points, this

power will be insufficient. A power level of +30 dBm EIRP gives improved performance and is

consistent with the Local Area Base Station specification with a higher gain (but still compact and

omnidirectional) antenna.

3.2 Interference amongst low power operators is manageable given

cooperation on mitigation techniques and conventions and a height limit

on outdoor deployments

Imposing separation distances between uncoordinated low power access points is not seen as a

practical solution for the co channel interference scenarios examined in this study. Instead

intelligent interference mitigation techniques available in LTE such as power control and dynamic

scheduling are thought to be more appropriate approaches for this band. Such techniques will allow

most users to access the whole channel bandwidth in most situations and for a graceful degradation

in bandwidth in limiting situations where the density of use and of cells is high. Technical

conventions to ensure these techniques are compatible amongst operators may be necessary to

maximise performance, but since the relevant technology is fast evolving and not standardised in

detail, it is recommended that this is left to licensees to coordinate amongst themselves.

Such interference techniques often amount to dynamic portioning of the spectrum amongst

neighbouring cells according to the signal conditions and demand. This partitioning will be more

successful given a wider bandwidth, and the associated user experience and capacity will relate

directly to bandwidth also, hence a 20 MHz channel is highly desirable.

A restriction on outdoor antenna height will help to reduce the range of interference from outdoor

cells. Interference ranges rise more rapidly as the antenna heights become significantly above the

heights of surrounding buildings. It is therefore recommended that the maximum height is set at or


a little above the typical height of residential buildings. At around 12 metres this would be

consistent with existing street furniture deployments by operators.

Technically, there is no specific limit on the number of operators licenced to use the shared access

channel. Interference mitigation techniques need to operate successfully given the limiting case of

an adjacent cell from another operator and this will occur even with two operators in the band. The

decision as to the number of licensees in the band is more of a policy decision. From a technical

viewpoint, we see no reason why 5 – 10 low power operators could not share the band. They will

however need to share the available capacity in overlapping deployments, typically in public areas,

strengthening the case for 2 x 20 MHz. In practice, when the number of operators is high it is likely

that operators will be incentivised to provide conditional roaming or share network equipment

access points in areas where deployments overlap rather than suffering degraded throughput from

interference.

3.3 An underlay approach presents interference and coordination

challenges, so we recommend a hybrid approach if dedicated 2x20 MHz

spectrum is not feasible

If the low-power spectrum block is provided on an underlay basis with a high power license, there

are several scenarios where interference between the systems can be significant. There exist

mitigation techniques to address all of these, but degradation of the performance may still be

significant in the absence of inter-operator roaming. It may also be complex to include the required

safeguards in the form of licence conditions. A dedicated spectrum approach is therefore preferred,

giving certainty to both forms of operator.

If however this option is not available with 2 x 20 MHz of spectrum, an intermediate solution would

be to overlap just half of the low power channel with a high power channel and introduce licence

conditions to ensure that interference is avoided. This may involve the high power systems always

take priority in the overlapping portion of the spectrum. This approach could provide a good balance

between the opportunities for both high power and low power licensees if appropriate licence

conditions can be arrived at.


3.4 Our recommendations rely on the use of interference mitigation

techniques and other coordination measures, facilitated by a code of

practice agreed amongst licensees

The separation distances to minimise interference at the proposed transmit power to ensure

reasonable coverage are not thought to be practical for expected deployments. However, no

separation distance is required if vendors implement intelligent scheduling and operators are

prepared to accept throughput degradations proportional to the number of uncoordinated

deployments in an area for the minority of users most affected by interference from neighbouring

cells.

A code of practice amongst operators making use of the shared band is recommended to ensure

capacity is maximised. This should include:

Implementing intelligent scheduling schemes that back off target throughputs and share capacity where neighbouring low power access points and, in the underlay case, macrocells are detected.

Implementing power control so that low power access points and associated UEs never transmit at higher powers than required for the target throughput.

Ensuring low power access points target low speed UEs so that the UE channel and it’s resource allocation isn’t changing frequently with time and can be tracked by other adjacent uncoordinated access points attempting to share capacity.

If separation distances are used to mitigate interference, the power spectral density (PSD) used on control channels should not exceed the PSD on the shared data channels. This is to ensure that the target SINR for control channels is achievable at the victim receiver if separation distances are set based on assumptions on the transmit power of the aggressor at a target throughput.

In the underlay or hybrid situation, priority should be given to the macrocell network with the low power access point reducing its power and accepting throughput degradations when a macrocell is detected.

Although the licencees should be able to choose the technology which they deploy, efficient

interference mitigation amongst operators is likely to require use of the same time and frequency

resource block sizes amongst operators, which may in practice lead to a convention on the use of a

single technology, most likely FDD LTE. and compatible with dynamic scheduling approaches and

environment measurements by both the UE and Home eNode B (HeNB) as outlined in the 3GPP

specifications for LTE.


3.5 We recommend positioning the shared channel at the upper end of the

FDD block, although further study is needed

By positioning low power devices at the upper end of the paired 2.6 GHz spectrum, an additional

frequency separation is introduced between high power macrocells and radar receivers above

2.7GHz. Although this study has not examined this case explicitly, this would seem helpful in

providing an additional ‘guard band’ between high power transmissions and the radar systems.

However, in some circumstances low power access points may have relaxed emission specifications,

which could create noise rise to radar receivers and we recommend that this situation is examined

explicitly.

The interference into low-power systems from radar systems and from TDD systems in the unpaired

2.6 GHz spectrum is essentially identical to that experience by high-power systems, so the choice of

spectrum arrangement to address this is not an essentially technical issue, but one relating to which

licensees could best tolerate such interference. It may be more feasible to increase the density of

low power cells in affected areas to mitigate this issue, but this option is equally available to high-

power licensees.

On balance, positioning the low-power shared channel at the top end of the FDD band seems a

sensible choice, but the case is not overwhelming.

3.6 Lessons learnt from the DECT guard band study and Wi-Fi

There are few precedents for the concurrent low-power use of spectrum amongst multiple

operators, and we are aware of none which relate to licenced use of a high speed mobile data

technology.

However, there are two comparable cases which may provide some useful lessons: firstly the Ofcom

award in 2006 of the 1781.1-1785 MHz and 1876.7 MHz – 1880 bands, commonly known as the

“DECT guard band” or “low power GSM band” and the widespread use of Wi-Fi technology in the 2.4

GHz band.

3.6.1 The DECT Guard Band

In May 2006 Ofcom awarded licences to 12 operators on a concurrent low-power basis. Although

the award was made on a technology neutral basis, the band is included in the definition of the DCS-

1800 MHz band supported by GSM mobiles, so GSM remains the most likely technology and it is

believed all commercial use of the spectrum has employed GSM technology. The technical

conditions in the licence included the following11:


A maximum outdoor transmitter antenna height of 10 metres above ground level

In the central 3 MHz of the band, a maximum EIRP of 0 dBm / kHz (para 9a applies) and 7 dBm/kHz (para 9b applies), corresponding to 23 dBm and 30 dBm in 200 kHz respectively.

Out of block emissions equivalent to those found in the GSM specification.

Ofcom required licensees to enter into a Code of Practice on Engineering Coordination within six months of the award.

The technical conditions were informed by a technical study12 conducted by Ofcom, some of the

conclusions of which were:

A low power system based on GSM pico cells operating at the 23dBm power level can provide coverage in an example multi-storey office scenario. Two pico cells per floor would meet the coverage requirements in the example 50m × 120m office building. For a population of 300 people per floor, the two pico cells would also meet the traffic demand.

Analysis of interference between neighbouring office buildings with indoor GSM pico cells operating on the same radio frequency indicates that a 97% probability of call success inside each office could be achieved with 550m separation between buildings if there were no obstructions between them. For a building separation of 150m the probability of call success is achieved is better than 90%. We conclude that coordination is necessary.

It is possible to serve users up to 40m within a building using an external base station with a power limit of 23dBm. However, to penetrate to users 50 meters within a building would require a higher power (30dBm would be needed to give a reasonable separation between the base station and building).

An outdoor micro cell could cause interference to an in-building pico cell system. At 3km a 23dBm micro cell reduces the call success probability on the pico cell system below 90% while a separation of 10km would be required for 97% call success. These figures are reduced significantly if there is an obstruction in the path. Adding a building on a 730m path gives a call success rate of 97%. However, if the outdoor micro cell has a power of 30dBm it is not possible to achieve a call success of 97% even with an obstructing building in the path unless the distance between the cells is unreasonably long.

We propose a maximum antenna height for outdoor installations of 10m as a means to reduce the occurrence of unobstructed interference paths. We also propose a maximum power level of 23dBm EIRP to prevent interference over a significant area.

A probabilistic analysis of interference between co-frequency indoor GSM pico cells located within a row of terraced houses indicates that a 97% probability of call success inside each house could be achieved with a separation of two houses.

A probabilistic analysis of interference between co-frequency indoor GSM pico cells located within houses in opposite terraces indicates that coordination will be required and that it will only be possible to assign frequencies from a total of 15 at random if the usage percentages are relatively low.

These act as a useful point of comparison with the conclusions from the present study. However in

making such a comparison it should be noted that there are significant differences between the

DECT guard band award and the potential 2.6 GHz low power award under consideration here:

Using GSM technology, the DECT guard band award allowed for at least 15 GSM carrier frequencies. Thus there was considerable scope for using frequency separation to avoid interference until the density of cells and the required capacity exceeded a significant level. In the current case, achieving the highest peak data rates requires that all cells are capable


of transmitting over the full bandwidth. Despite this, LTE and other next-generation mobile systems are capable of dynamically adjusting the bandwidth occupied according to the demand from users and the interference levels in different parts of the channel occupied, so that a form of dynamic frequency reuse can be applied, at the expense of a reduction in capacity and user throughput.

The automated interference sensing and mitigation techniques which are now standard in femtocells had not been widely envisaged at the time of the award, so it was not possible to build in these capabilities as an assumption in setting the technical parameters in the DECT guard band award.

The use of GSM technology suggested a mainly connection-oriented protocol, based particularly on voice services. In such systems, relatively short instances of interference can cause dropped calls to the major detriment of perceived quality. By contrast, LTE and similar systems are entirely packet-oriented and targeted mainly at data applications, where a temporary reduction in throughput may not be noticeable to a user and may be entirely overcome by appropriate prioritisation of packets according to the required quality of service.

LTE and similar technologies achieve a far higher spectrum efficiency than GSM, so they are capable of delivering a far higher level of traffic for equivalent signal conditions and bandwidth.

The 2.6 GHz band produces higher propagation losses for a given distance than the 1.8 GHz band, so smaller distances should give equivalent protection levels.

These issues, taken together, suggest that the technical conditions applicable to the potential new

award should be no more restrictive than those of the DECT guard band and that some of the

conditions could potentially be relaxed to some extent, including potentially the transmit power

levels, the antenna heights and the need for coordination.

3.6.2 Wi-Fi in the 2.4 GHz band

The use of Wi-Fi technology in the 2.4 GHz band is governed by a very different regime to that of the

potential 2.6 GHz low power allocation. However there are some interesting similarities which may

inform the current considerations.

Wi-Fi at 2.4 GHz is formally a system based on the IEEE 802.11b, g and n specifications. In the UK,

the main band of operation is 2.4 – 2.4835 GHz, although there is increasing use of Wi-Fi in the 5

GHz range. Within this band the technical conditions are set out by a Radio Interface Requirement

200513. The technical conditions are simple, being mainly a maximum transmit power of 20 dBm

EIRP. Provided equipment complies with this Interface Requirement, it may be operated on a

licence-exempt basis. Furthermore, since July 2002, commercial services have been permitted in

this band14. A vast number of devices now operate in the band, including commercial services

operated on a large scale by operators such as BT, who operate over 2 million UK Wi-Fi hotspots15,


and The Cloud who operate more than 22,000 access points across Europe16. In January 2011 O2

announced plans to deliver a network which will be “at least double the number of premium

hotspots offered by BT Openzone and The Cloud combined by 2012.”17 This neglects the vast

number of individual private deployments in homes and offices.

There are a number of technical factors which might suggest that services over Wi-Fi would yield

poor performance:

A complete lack of coordination amongst deployments.

The presence of both outdoor and indoor deployments with no constraints on height.

Many other devices which use the same spectrum band, including Bluetooth headsets, medical and scientific devices, baby alarms and even microwave ovens.

A high likelihood of adjacent devices operating on the same radio frequency. Each Wi-Fi transmission is at least 22 MHz in bandwidth, so there are only three non-overlapping channels. Although newer Wi-Fi systems incorporate interference sensing and automated retuning, many existing devices do not. Indeed devices from the same manufacturer all typically come pre-configured with the same frequency channel.

No power control in existing devices: although newer variants of Wi-Fi do include some power control, this is rare in existing devices, so both the access points and the client devices usually radiate at a fixed power of around 100 mW.

Indeed, in 1999, before Wi-Fi in its current form gained commercial acceptance, a study for Ofcom’s

predecessor regulator the Radiocommunications Agency concluded that18:

“Operation of high performance telecommunication networks in a very dense urban

environment such as the City of London "square mile", which is also subject to a relatively high

number of OBTV transmission, is unlikely to be viable. Operation of a single such network in

other more typical urban areas should be viable in most instances, providing due account is

taken of the projected future increase in interference levels. ”

Further, a study by Mass Consultants Ltd. on behalf of Ofcom19 did identify real performance

challenges with densely deployed Wi-Fi systems deployed in locations such as the centre of London. In such

locations, the report found that it is rare for the user data frame rate to exceed 10% of the total frame rate,

so that existing Wi-Fi protocols were using a substantial part of the 2.4 GHz band without carrying significant

quantities of user-generated content. In addition this study found that in many situations where Wi-Fi

performance was degraded, and originally attributed to congestion, the interference was due to other non

Wi-Fi friendly devices such as baby alarms and CCTV cameras making use of the band.

Nevertheless, for most users, in most locations and for most of the time Wi-Fi delivers a useful

service, with data rates which support a wide range of services. When Wi-Fi systems encounter interference


on a particular packet, they retransmit it a random period of time later, so that the throughput is degraded

but data still flows. Although there are various ways in which the original Wi-Fi protocols can be improved

(and are being improved in newer versions), they still provide a satisfying service in most cases. Significant

value has been ascribed to the use of Wi-Fi: for example, one study suggests that Wi-Fi usage in the home,

for only the purpose of broadband extension, may be generating anywhere between $4.3 and $12.6 billion

in annual economic value for consumers in the United States20.

Comparing Wi-Fi with the use of LTE (or a similar technology) in the potential low-power concurrent

allocations, there are several reasons to expect a service which is superior to Wi-Fi:

A limited number of operators, with the potential to monitor and coordinate deployments via co-operation amongst themselves and by the deployed devices reporting their locations and parameters to a central controller for each operator and potentially amongst operators.

A managed protocol in LTE which delivers assured quality-of-service streams which can be differentiated according to device and service to make best use of the available signal quality.

A greater opportunity for operators to agree consistent technologies and interference mitigation conventions, avoiding the risk of interference from dissimilar devices.

Uplink power control to minimise interference.

A range of interference mitigation techniques, comprising the whole range of techniques which have already been tested in commercial operation for 3G femtocells and which are supported in LTE standards, including downlink transmit power control and both coordinated and distributed resource scheduling techniques.

Support for full mobility amongst cells with seamless handovers even at high speed.

Ability for the same devices to handover to wide area, high power macrocells given appropriate roaming arrangements amongst operators.

The only significant downside relative to Wi-Fi is the relatively small spectrum bandwidth available (20 MHz

or 40 MHz in total compared with 83.5 MHz). On balance, then, it should be expected that low power

concurrent licenced operation could provide a service which is beyond the level of performance of Wi-Fi.

Given the proliferation of such systems, they could potentially act as a complement or even a replacement

for Wi-Fi, such as in cases where a high level of service assurance is required for business-critical

applications: a form of “grown-up Wi-Fi”.


4 Our assumptions on LTE devices and interference mitigation

techniques

4.1 LTE device parameters

We have made assumptions for LTE devices based on the 3GPP standards, related technical studies

and previous work published by Ofcom. The parameter assumptions were agreed with Ofcom prior

to the modelling work.

Table 1:LTE Device Parameter Assumptions: User Equipments

Units Value Source

Max Transmit

Power dBm 23

TS 36.101

Also matches 3GPP R4-092042

Antenna Gain dBi 0

Assumption in line with 2G Liberalisation

Also matches 3GPP R4-092042

Antenna Height m

0.5 from

floor

Cable, Combiner

and Connector

Losses dB 0 Assumption in line with 2G Liberalisation

Receiver Noise

Figure dB 9 Assumption in line with 3GPP R4-092042

Adjacent Channel

Rejection Ratio dB

31.5dB for

10MHz and

below

28.5dB for

15MHz 3GPP TS 36.101


25.5dB for

20MHz

Body loss dB 0 Assumption in line with 2G Liberalisation for data-centric devices

Transmit power

control Dynamic

range dB 64 TR36.942

Table 2: LTE Home Local Area Base Station Parameter Assumptions

` Value Units Source

Max Transmit

Power 24 dBm

As per maximum transmit power for a local area

basestation given in TS 36.104. Also matches with

stakeholder comments.

Antenna Gain 5 dBi

Assumption of a small directivity value achieved by

suppressing emissions in the vertical direction, R4-

092042

Antenna Height

5 (default), up to

15m m Location on a lamp-post or other pole

Cable, Combiner

and Connector

Losses 0 dB

Assumption in line with 2G Liberalisation for a

tower-mounted device

Adjacent channel

rejection ratio 43.5 dB TS 36.104 section 7.5.1

Receiver Noise

Figure 8 dB

Assumption in line with the noise figure used for

HeNBs in 3GPP R4-092042.

TS 36.101 section 7.2 specifies the same receiver


sensitivity for both home basestations and local

area basestations so we are assuming that the

noise figure for this outdoor low power eNodeB

(i.e. a local basestation) will be the same as the

home eNode B case.

Table 3: LTE Home Home e Node-B Parameter Assumptions

Value Units Source

Max Transmit

Power

20 for residential

environment

24 for public

environment and

enterprise dBm

For a residential environment assess a

power range from 0-20dBm as per R4-

092042 and the max power limit for a

home BS in TS 36.104

For a public area environment assess a

power range from 0-24dBm as per the

max transmit power for a local BS in TS

36.104

Antenna Gain

0 for residential

3 – 5 for public

area and

enterprise dBi Assumption in line with 2G Liberalisation

Antenna Height

2.5m from the

floor m Ceiling fit or wall mount

Cable, Combiner

and Connector

Losses 0 dB Assumption in line with 2G Liberalisation

Adjacent channel 43.5 dB TS 36.104 section 7.5.1


rejection ratio

Receiver Noise

Figure 8 dB Assumption in line with 3GPP R4-092042

Table 4: LTE Macrocell Parameter Assumptions

Value Units Source

Downtilit 6 deg

Assumption in line with 2G Liberalisation and 2.6GHz

and Radar Study

EIRP (10 or 20MHz

channel) 60 dBm

From TS 36.104 the maximum transmit power set is

independent of bandwidth choice so the EIRP is the

same for 10 and 20MHz deployments.

Table 2 of 3GPP R4-092042 gives BS transmit power of

46dBm for macrocells and an antenna gain of 14dBi

giving an EIRP of 60dBm.

Antenna height 25 m

Assumption in line with 2G Liberalisation and 2.6GHz

and Radar Study

Antenna gain 14 dBi Assumption in line with 3GPP R4-092042

Max att due to

patterns 20 dB Assumption in line with 3GPP R4-092042

HPBWhor 70 deg Assumption in line with 3GPP R4-092042

Adjacent channel

rejection ratio 43.5 dB TS 36.104 section 7.5.1


BS noise figure 5 dB 3GPP R4-092042

4.2 Quality of service and interference criteria assumptions

TBC

Target data rates focused on. Assume low power high density access points are for data hot spots

deployed for capacity benefits rather than to fill coverage gaps and so focus on upper end data

rates.

Coverage assumption of 90% of area which is 78% confidence level at the cell edge when adding

fading margins. Have included this in interference analysis so that the target SINR includes ensuring

that a 78% coverage confidence at the target degraded throughput at the cell edge if that cell edge

user is the single user in the cell and gets max available resources.

Interference criteria is difficult to set and stakeholder discussions didn’t reflect an industry standard

unified view on a target

4.3 Typical building sizes in target deployments

TBC

Types of environments include:

Residential

Office

Public areas – we’ve focused on the example of a shopping centre as it’s a challenging environment due to large area not only due to the large number of wall losses compared to train stations or airports with large open areas.

Business park / campus

Residential street coverage For each of these go through our assumptions on wall separation, average building sizes, wall losses

and building penetration. Coverage on multiple floors.

4.4 Propagation models used

In order to select appropriate propagation models for this study, we have first identified the

propagation path types which occur in the scenarios identified for analysis. A survey of models used


in other studies of interference amongst femtocells is summarised. Finally models are assigned to

each of the relevant path types by applying our judgement based on other studies and our

experience in modelling similar systems. The individual scenario descriptions in chapters 4 and 0

then identify the models used in each case.

While propagation models for specific scenarios can be highly accurate, we seek in this study to

adopt a more generalised approach, which typically involves the use of models such as those

provided by the ITU-R based on simple parameters such as distance, frequency and antenna heights.

The wide variety of geometries and materials used in buildings make such generalisation especially

difficult for the short range and indoor systems which are the subject of this study. In particular,

such generalised models are unlikely to provide an accurate assessment of system performance. For

our purposes this is acceptable since we seek to determine regulatory limits which protect against

interference in relatively extreme scenarios. However, future operators of these systems are likely to

need to adopt a more detailed approach to determine the level of service they will be able to offer

to their customers in particular situations.

4.4.1 Path types to be modelled

Depending on the scenario, a wide variety of differeing wanted and interfering signal paths need to

be modelled. The propagation models assigned to each of these may need to be rather different, so

we explicitly identify the path types here:

a) Outdoor, relatively short range for coverage, with rather low antenna heights, typically in the range

of 5-10m and with relatively unobstructed paths.

b) Outdoor interference paths from cells mounted with higher antenna heights, often above the roofs

of surrounding buildings.

c) Outdoor to indoor, where for example an outdoor cell is used to provide indoor coverage.

d) Indoor to indoor based on a low power cell providing coverage to an indoor user.

e) Indoor to outdoor interference, for example for an indoor low power cell creating interference to

outdoor users.

f) Indoor to indoor between buildings, for example for a low power cell creating interference to an

indoor user in another building.

g) Macrocell to indoor or outdoor users for determining interference arising from underlay use of low-

power cells.

4.4.2 Models used in other studies

The Ofcom technical study21 in support of the award of the 1781.1-1785 MHz and 1876.7 MHz –

1880 bands (“DECT guard band” / “low power GSM band”) makes extensive use of the propagation

models in the ITU-R recommendation P.123825 together with wall loss values as given in the COST

23124 project report. It uses the free space loss plus appropriate wall losses for modelling


interference between adjacent buildings. For calculating the loss between buildings with another

building in between, a diffraction loss was added based on ITU-R recommendation P.52622.

A 3GPP document R4-09204223contains common simulation assumptions specifically for modelling

FDD HeNB (LTE femtocells). This is applied within the main 3GPP technical report on interference

mitigation techniques for HeNB, TR 36.92128. The document recommends use of the following

propagation models for the median path loss at a given separation distance:

For UE to HeNB paths in the same house or apartment, two potential models are recommended. The main one corresponds to the COST231 Multi-Wall Model24 with parameters chosen as recommended in the original report for 1800 MHz when predicting for the 2 GHz band, and with an assumed 4.9 m separation between walls. The other model does not model walls explicitly, but accounts for them via an increase path loss exponent relative to free space loss.

For UE to macrocell paths, the following model is used: L (dB) =15.3 + 37.6log10R + Low

where Low is the loss associated with the outside wall of a building. The source for the model is not given but it follows the Okumura-Hata model for particular parameters.

Shadowing is modelled via a lognormal distribution with a standard deviation (location variability) of 8dB for macrocell paths, 4 dB with explicit wall loss modelling (the COST231 Multi Wall Model) and 10 dB with the simplified path loss model.

When paths between adjacent apartments are calculated, the same models are applied but an additional 5dB loss is applied to account for the loss associated with the wall separating the apartments.

The Femto Forum white paper on interference management in OFDMA femtocells30 follows these

recommendations closely.

In the Femto Forum white paper on interference management for UMTS27, the following models are

used:

ITU-R recommendation P.123825 for indoor to indoor paths.

ITU-R recommendation P.141126 for femtocell UEs interfering with macrocells. An additional wall loss component is included for UEs which are inside buildings.

Free space loss for calculation of ‘dead zone’ effects when UEs are very close to femtocells, with a minimum path loss limit applied corresponding to 1m separation.

The COST 231 – Hata model24 for macrocells, with an additional wall loss for paths to indoor UEs.

A 3GPP micro urban model for an apartment block-to-outdoor scenario. The reference for this model is not given.


4.4.3 Selection of models according to path type

Table 5 provides our selection of propagation models according to the path type. Further detail of

these models is provided in Annex A with the specific parameter values used. In many cases existing

published propagation models do not explicitly cover the 2.6 GHz band, necessitating corrections

which we have assigned based on the frequency trend of the model and other work previously

published by Ofcom.

Table 5: Selection of propagation models according to path type

Path Type Model

a) Outdoor, relatively short range for coverage, low height (5-10m).

The ITU-R P.141126 model is applied for line-of-sight

situations. The model provides lower bound and upper

bound losses, so use the average of these models (in

decibels).

b) Outdoor interference range (diffraction over building rooftops)

We apply the version of the COST 231 Walfisch-Ikegami

model24 which is embodied in ITU-R P.1411 §4.226, which

covers non line-of-sight situations and base station heights

4-50m. Although this is only recommended up to 1km by

ITU-R, the underlying model is valid up to 5km.

A building entry loss is added based on the COST231 non

line-of-sight penetration loss model in §4.6.3 of 24.

c) Outdoor to indoor (microcell providing indoor coverage)

The ITU-R P.1411 §4.226 model is applied, with the addition

of the penetration loss portion of the COST 231 LOS

building penetration loss model in §4.6.3 of 24.

d) Indoor to indoor (femto coverage)

For houses and indoor public areas:

The COST 231 Multi Wall Model in §4.7.2 of 24 is applied,

with a wall separation appropriate to the environment.

For indoor office environments:

The ITU-R P.1238 model25 is applied.

e) Indoor to outdoor interference (femto to

The COST 231 LOS model in §4.6.3 of 24 is used to model the


outdoor interference) building loss.

f) Indoor to indoor between buildings

The ITU-R P.1411 street canyon model26 is used with two

instances of penetration loss terms. However there is also a

case for including FSL as a sensitivity case.

g) Macro to indoor, macro to outdoor to determine underlay interference

For these cases there is no need to model the loss explicitly

as the calculation are conducted by assuming a specific

signal level applicable to the macrocell user on the edge of

coverage.

4.5 Small cell interference mitigation techniques

In conventional mobile networks, cells are carefully planned and optimised by skilled engineers,

using a variety of prediction, measurement and analysis tools. In a small cell deployment such an

approach may not be viable, both because of the relatively higher number of cells involved and

because some of the cells – especially femtocells for homes and small offices – may be deployed in

locations which are not under the direct control of the operator.

In order to overcome this, manufacturers and standards bodies have investigated and implemented

a wide range of techniques for determining the interference and coverage situation in which a small

cell finds itself, and adjusting its parameters dynamically within ranges set by the operator to meet

defined service quality targets. The parameters adjusted include the transmit powers of both the

cells and the mobile devices which connect to them, the carrier frequencies and codes used by the

cells, and the specific time and frequency resources used by adjacent cells for particular users. All of

these parameters can be adjusted in response to a range of measurements, including location and

performance data from the cells themselves (which may monitor both uplink and downlink

frequencies), from mobiles and from the wider operator network and management system. In 3G

systems, such techniques have been studied in depth 27,28 and are reported by operators to be

effective in field operation, including when deployed co-channel with macrocells. For example, AT&T

have said29:

“We have deployed femtocells co-carrier with both the hopping channels for GSM macrocells and with

UMTS macrocells. Interference isn’t a problem. We have tested femtocells extensively in real customer

deployments of many thousands of femtocells, and we find that the mitigation techniques implemented

successfully minimise and avoid interference.The more femtocells you deploy, the more uplink

interference is reduced”


However, the specific techniques used to mitigate interference and manage performance are not

generally mandated by standards, but are left to the individual manufacturer and operator to

determine. The specifications provide the ‘hooks’ to enable these techniques to operate

successfully. For 3G, the following key dimensions were identifiedin 27:

“Femtocell downlink power – if femtocells transmit inappropriately loudly, then the cell may be large, but non‐members of the closed user group will experience a loss of service close to the femtocell. On the other hand, if the femtocell transmits too softly, then non‐group members will be unaffected, but the femtocell coverage area will be too small to give benefit to its users.

Femtocell receiver gain – since UEs have a minimum transmit power below which they cannot operate, and since they can approach the femtocell far more closely than they can a normal macrocell, we must reduce the femtocell receiver gain, so that nearby UEs do not overload it. This must be done dynamically, so that distant UEs are not transmitting at high power, and contributing to macro network noise rise on a permanent basis.

UE uplink power – since UEs transmitting widely at high power can generate unacceptable noise rise interference in the macro network, we signal a maximum power to the UE (a power cap) to ensure that it hands off to the macro network in good time, rather than transmit at too high a power in clinging to the femtocell.”

When such mitigation techniques are applied, there is potential for very high capacity densities

arising from intensive re-use of spectrum over a given area and in 27 simulations indicate the

potential for air interface data capacity to increase by over a hundredfold by the introduction of

femtocells co-channel with macrocells. Our report for Ofcom on 4G Capacity gains identified the

potential for such small cells to work in conjunction with the spectrum efficiency gains available

from 4G technologies to deliver sufficient capacity for some of the highest potential levels of

forecast traffic over the next ten years6.

However, the scenario under investigation in this study is rather different from the one which is

usually studied when evaluating such techniques. Differences occur in the following ways:

Interfering cells may not be under the control of the same operator, so there is less opportunity for coordinated optimisation. This situation could be reduced by setting appropriate limits and conventions on cell behaviours, either via licence conditions or more likely via an agreement amongst operators.

Users belonging to one operator may come close to the low-power cell of another and both suffer and cause interference as result (the so-called ‘visitor problem’). This situation also arises in a single operator situation when closed subscriber groups are used, but may be more common when so many operators share the same channel and is exacerbated in the case of a potential underlay network. A possible remedy for this would be for operators to allow roaming of users amongst cells, potentially only in cases of extreme interference, similar to the hybrid access mode which is already envisaged for use by individual operators.

If low-power operators wish to deliver the full potential bitrate which the technologies can offer, they will need to occupy the full 10 MHz or 20 MHz channel bandwidth, giving no opportunity to shift carrier frequencies when extreme cases of interference are encountered. However, both LTE


and WiMAX utilise OFDMA technology which allows subcarriers within the bandwidth to be allocated to particular users according to their interference conditions.

Two key studies of interference mitigation techniques for OFDMA systems are summarised here.

4.5.1 Control Channel Interference in LTE

The modelling of interference in section Error! Reference source not found. focuses on the

throughput performance of user data. However, LTE also requires successful decoding of control

channels. This section explains the issues associated with these as background to the discussion of

both data and control channel mitigation techniques in the following sections.

In shared data channels, resources can be dynamically scheduled to be allocated around the

interference. Scheduling is slightly less flexible on the uplink as resource blocks can only be assigned

in contiguous assignments whereas on the downlink allocations can be distributed, but the same

general principles apply.

However, common control channels are less flexible as they have fixed time and frequency resource

locations, as illustrated in Figure 2.

Figure 2: Control channel allocation in LTE downlink

© 2011 Real Wireless Ltd. 1

PDCCH

Ant 2 ref

Ant 1 ref

PCFICH

Ant 2 ref

PDSCH Sec Sync Prim Sync

Ant 1 ref

PCFICH

PCFICH

PCFICH

Ant 1 ref

Ant 1 ref

Ant 2 ref

Ant 2 ref

PDCCH

PDCCH

PDCCH

PDCCH

PDCCH

PDCCH

PDCCH

PDCCH

PDCCH

PDCCH

PDCCH

PDCCH

PDCCH

PDCCH

PDCCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

Sec Sync

Sec Sync

Sec Sync

Sec Sync

Sec Sync

Sec Sync

Sec Sync

Sec Sync

Sec Sync

Sec Sync

Sec Sync

Prim Sync

Prim Sync

Prim Sync

Prim Sync

Prim Sync

Prim Sync

Prim Sync

Prim Sync

Prim Sync

Prim Sync

Prim Sync

PBCH

Ant 2 ref

Ant 1 ref

Ant 2 ref

Ant 1 ref

Ant 1 ref

Ant 1 ref

Ant 2 ref

Ant 2 ref

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PBCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

PDSCH

Su

b c

arr

iers

1 slot (7 OFDM symbols) 1 slot (7 OFDM symbols)

1ms subframe

Downlink control channel with resource allocations for the UE transmitted on first 1-2 symbols every subframeacross all sub carriers

PCFICH gives length of PDCCH in time. Treat with the PDCCH

Synchronisation channels transmitted twice per 10ms frame on central 72 sub carriers

Broadcast channels transmitted once per 10ms frame on central 72 subcarriers


The mitigation techniques discussed in the subsequent sections address this issue, but there are also

potential mitigation approaches which are somewhat specific to the case of concurrent low power

spectrum:

• LTE supports multiple bandwidths, so the 10MHz or 20 MHz bandwidth could be split into a number

of smaller BWs per operator so that the control channels were not overlapping, at the expense of a

loss of capacity and flexibility.

• The hybrid spectrum approach proposed in section *prevents overlapping control channels with the

macro cell but relies on spectrum aggregation from release 10 to make use of the overlapped

portion of spectrum by the low power devices.

4.5.2 Femto Forum study of interference management for OFDMA femtocells

The Femto Forum published a study “Interference Management in OFDMA Femtocells”30, focusing

entirely on the case of closed user groups, which is comparable with the case of multiple operators

with no roaming arrangements between them. The study focused on 2GHz, but the general findings

are expected to be reasonably applicable to 2.6 GHz. Ten key scenarios where identified where

interference could be challenging. Only two of these related to interference amongst femtocells and

the associated mobiles, while the others were related to interactions between femtocells and

macrocells. Note that all femtocells in this study were deployed indoors.

Scenario Outcome

Macrocell Downlink Interference to

the Femtocell UE Receiver (A1)

Throughput degradation was limited to less than 1%

Macrocell Uplink Interference to the Femtocell Receiver (B1)

Interference power lower than thermal noise power: no

detectable degradation of average throughput.

Femtocell Traffic Channel Downlink

Interference to the Macrocell UE

Receiver (C1)

Degradation can occur without mitigations , but mitigations

are effective in reducing degradation – see below.

Femtocell Control Channel Downlink

Interference to the Macrocell UE

Significant degradation of performance can be experienced

for a visiting (same building) macrocell UE when within a


Receiver (C2) ‘deadzone’ of tens of meters depending on the transmit

power without mitigations. A ‘passing’ UE protected by a

building wall experiences interference over several meters

from the femtocells. Power control is expected to be

effective in a similar way to the previous scenario but is not

analysed in the report. Handover of the macrocell UE to

another frequency is another mitigation suggested.

Femtocell Uplink Interference to the Macrocell NodeB Receiver (D2)

Potential interference is mitigated by placing a power cap on

femtocells UEs as function of pathloss to the macrocell,

based on minimising noise rise below a specified level. If the

target is less than 0.2 dB noise rise, macrocell average sector

throughput is degraded by between 0 and around 20% and 5-

percentile user throughput by up to 40% as the number of

femtocells increases from 0 to 80 per sector. The throughput

degradation of the femtocells users is significantly less.

Although the total throughput of both systems combined is

always increased by the addition of the femtocells, this is not

so relevant in the case of different operators.

A more advanced scheme signals the interference actually

experienced at the macrocell via an interface between the

two systems (the X2 interface defined by 3GPP) and

significantly reduces the macrocell degradation, but would

require close cooperation between the macrocell and low-

power operators.

Femtocell Downlink Interference to Nearby Femtocell UE Receivers (E)

A distributed fractional frequency reuse system is proposed

to mitigate interference in this situation. Each cell constructs

a neighbour list through network listening and user reporting.

This is used to establish a reuse pattern of the available

bandwidth to maximise performance. Performance of the

worst-affected is substantially improved relative to having no

reuse scheme, with 20-30% of users achieving double the

throughput depending on the penetration of femtocells. This


would rely on the operators all adopting a similar scheme.

A dynamic interference avoidance scheme is also studied,

which can be achieved via over-the-air coordination to avoid

a physical interface between operators. It is found to

significantly improve performance over uncoordinated

approaches, especially as regards the latency experienced in

the presence of strong interference from loaded cells.

Femtocell Uplink Interference to Nearby Femtocell Receivers (F1)

Full mobile transmit power and two adaptive power control

mitigation techniques (fractional and noise rise) are

compared. At large deployment densities, fractional power

control achieves better cell edge throughput than full power

at the cost of lower average cell throughput. At low

deployment densities, allowing full power operation is found

to give higher system performance. The probability of large

degradations is anyway found to be low such that overall

system performance is acceptable.

Macrocell Downlink Interference to an adjacent channel Femtocell Receiver data channel (G3)

A 5 MHz adjacent channel frequency separation is examined.

With 3GPP adjacent channel selectivity levels, 99% of

locations experience SINR of 8dB or greater with 0dBm

femtocell power. Overall performance degradation should

therefore be small provided the femtocells power is set

sufficiently high to provide good quality coverage in a given

environment.

Macrocell Downlink Interference to the adjacent channel UE Femtocell Receiver (control channel) (G4)

Similar findings to the previous scenario – no major

degradation.

Macrocell Uplink Interference to the adjacent channel Femtocell Receiver (H1)

Femtocell UEs are assumed to increase their power in

response to the approach of a high power macrocell UE. 95%

of locations are found to experience an SIR greater than

14dB, so the impact should be small.


In the case of potential downlink interference from femtocells to macrocell user equipments (UEs),

performance degradation can be significant if unmitigated. Three potential interference mitigation

techniques were studied:

a) Distance based power control, where the femtocell transmit power is reduced at close distances

from a macrocell. This requires that the locations of the macrocells are known to the femtocell

operator, which is unlikely in our scenario. The cell throughput was found to be largely unaffected at

small numbers of femtocells, to reduce throughput degradation at higher numbers, but eventually to

lead to significant throughput degradation at higher densities. The overall aggregate cell throughput

was largely unaffected, but there are significant throughput degradations for a small number of

particular users.

b) Power control based on pathloss. In this case where a power limit is placed on femtocells to

protect the macro downlink as a function of the measured pathloss to the neighbouring macrocells.

The results indicated that interference was well controlled, with an increased average throughput

and reduced outage probability even compared with the macrocell-only case, although clearly in this

situation macrocell UEs are able to access the femtocells.

c) Power control based on pathloss and detection of the presence of victim UEs. This can be achieved

without knowledge of the macrocell locations. Additionally the femtocell detects transmissions from

the macrocell users and adjusts its power accordingly. The technique appeared effective and

increased the throughput available to UEs.

In summary, interference mitigation techniques are important to achieve good performance. This is

particularly acute when considering interference to co-channel macrocell users, where even with

mitigation techniques, the macrocell operator’s performance can be significantly affected, despite

overall (macro+femto) capacity being increased. Amongst shared low power operators, good

performance appears possible and can be enhanced via cooperation on the forms of interference

mitigation technique which are implemented.

4.5.3 3GPP Technical report

3GPP technical reports do not form mandatory specifications but are purely informative.

Nevertheless technical report TR 36.92131 provides details of the means by which various factors can

be controlled to mitigate interference, particularly the transmit power but also including variable


bandwidth and allocation of subcarriers. It also provides guidance to operators on techniques which

may be used to set the appropriate values for these factors.

Potential measurements which can be made by a HeNB (LTE femtocells) include those specified

below:


Measurements of Measurement Type Purpose Measurement Source(s)

All cells

Received Interference Power Calculation of UL interference towards HeNB (from MUE)

HeNB UL Receiver

Surrounding cell layers Cell reselection priority

information Distinction between cell types based on frequency layer priority

HeNB DL Receiver

CSG status and ID Distinction between cell layers based on CSG, and self-construction of neighbour list,

HeNB DL Receiver

Macrocell layer

Co-channel RSRP

Calculation of co-channel DL interference towards macro UEs (from HeNB) Calculation of co-channel UL interference towards macro layer (from HUEs) Calculation of co-channel UL interference towards HeNB (from MUEs) based on estimated MUE Tx power Determine coverage of macro cell (for optimization of hybrid cell configuration)

HeNB DL Receiver HUE MUE (in case of hybrid cell)

Co-channel RSRQ Determine quality of macro cell (for optimization of hybrid cell configuration)

HeNB DL Receiver HUE MUE (in case of hybrid cell)

Reference Signal Transmission Power

Estimation of path loss from HUE to MeNB

HeNB DL Receiver

Physical + Global Cell ID

Allow HeNB to Instruct UEs to measure specific cells. Allow UE to report discovered cells to HeNB.

HeNB DL Receiver HUE

Detection of UL RS Detection of victim UE HeNB UL Receiver

Co-channel received CRS Êc (measured in dBm)

Measurement is used to determine whether HeNB is close to dominant Macro cell, or whether it is close to macro-cell-edge border.

HeNB DL Receiver

Adjacent HeNBs

Co-channel RSRP

Calculation of co-channel DL interference towards neighbour HUEs (from HeNB) Calculation of co-channel UL interference towards neighbour HeNBs (from HUEs)


Reference Signal Transmission Power

Estimation of path loss from HUE to HeNB

HeNB DL Receiver

Physical + Global Cell ID Allow HeNB to Instruct UEs to measure specific cells Allow UE to report discovered cells to HeNB.


The report lists techniques for controlling interference for both data and control channels in both

the downlink and the uplink.

Downlink


Data channels

Frequency partitioning to control downlink H-NB to macro eNB data channels

HeNBs operating in underlay mode could obtain frequency partitioning information from the

overlayed eNBs if they have a downlink receiver. This does not require that the RBs used by the

macrocell are completely avoided as this was cause a severe loss in capacity. Instead the macro eNB

can preferentially use one set of resource blocks for users close to the macro and others for cell

edge users. Based on a knowledge of its location or of the path loss to the macro, the HeNB can

avoid resource blocks likely to be used by nearby macro UEs.

Control of HeNB downlink interference among neighbouring HeNBs

LTE supports subband fractional frequency reuse via channel quality information reporting. In a 10

MHz system with 6 RBs/subband, there are 8 regular subbands and one short subband that could be

used to implement fractional frequency reuse.

The HeNBs can listen to neighbouring control channel and reference signal transmissions, determine

the corresponding cell IDs and measure the associated path losses. UE measurement reports can

also be used. Based on these measurements HeNBs can use a fractional frequency reuse scheme to

to avoid transmitting on the same resources as neighbouring HeNBs. This could be achieved via a

centralised coordinator which uses the reports to form an ‘adjacency graph’ based on the

measurement results and assigning resources to the HeNBs accordingly. This would require a specific

interface between low power operators, or perhaps to a third=party which would operate the

central controller on behalf of the operators. Alternatively each HeNB could construct its own

‘jamming graph;’ and one of a number of known distributed algorithms could be used to select

resources. For example, an autonomous technique described in 32 concludes that:

“Extensive simulation results provide evidence that the presented concept renders average cell throughput virtually

insensitive to the density of neighboring femtocells, without compromising cell edge user throughput when

compared to universal frequency reuse. Hence, it provides a fully distributed (scalable) and self-adjusting

frequency reuse mechanism, which allows for uncoordinated eNB deployment without prior (expensive and

manual) network planning.”

This would rely on all low power operators adopting a similar technique, or at least operating in a

‘fair’ manner. The algorithms used on the HeNBs would not necessarily need to be identical, but

they would need to be used in a manner which still led to fairness. This requires less technical

coordination and is perhaps more likely for the low power concurrent operators than an approach

involving a specific interface and a central controller.


Control of HeNB downlink interference by dynamically changing closed subscriber group IDs

When closed subscriber groups (CSG) are used, interference can occur to UEs close to an HeNB

whose CSG it does not belong to. This would be the normal case for UEs of low-power operators.

One approach to minimising this is for a special dedicated CSG to be assigned and for the CSG IDs for

HeNBs to be assigned to the special CSG ID in a coordinated manner, reducing the number of

mobiles experiencing interference. This is essentially a limited form of roaming arrangement, to be

used only when interference is experienced or predicted. The report indicates that the control of this

process would be conducted by a centralized controller, but it is possible to envisage distributed

approaches to the same technique.

Victim UE aware downlink interference management

Byexplicitly detecting the presence of nearby UEs liable to be victims of interference, it is possible to

ensure that mitigation techniques such as power control and resource allocation are only used when

required, avoiding wasted capacity. This can be done on the basis of reported measurements from

the UEs (which may require access to an appropriate interface between cells) or based on detected

uplink transmission, which can be done without access to an interface. UEs likely to suffer

interference are also likely to be transmitting at high power and close to the aggressor cell, so these

transmissions should be detectable with high reliability.

Power control

Power control is an important interference mitigation technique, allowing a careful trade0off

between coverage and interference on a dynamic basis. HeNBs will typically include a downlink

receiver, allowing them to detect surrounding cells at their location. This alone may not, however,

provide a good indication of interference conditions, since the victim UEs are in entirely different

locations and could suffer significantly higher interference (Figure 3 provides an example).

Measurements from UEs are therefore also useful in setting the power appropriately. The setting of

the power based on these measurements depends on an appropriate tradeoff between interference

caused and the performance degradation in your own cell. Low-power operators may wish to agree

on some of the relevant parameters in order to ensure a fair approach.

The quality of GPS signals can also be used as an input to the HeNB power control process. Poor GPS

detection performance (based on the number of satellites detected and the reception quality) is

correlated with scenarios where the HeNB is indoors or otherwise shielded and relatively unlikely to

cause interference to macrocells. When GPS performance is good, the HeNB can set a lower power

to protect the macrocells.


Figure 3: Scenarios where HeNB transmit power are not appropriately set by HeNB measurements alone

(Source: 3GPP, from 28

)

Control channels

Time shifting at symbol level

Control channels must be decoded successfully for any data to be exchanged, so they have

to be received successfully. In LTE the main control channels (SCH/PBCH) occupy several

resource blockscentred within the channel bandwidth and repeating periodically. If base

stations are unsynchronised in time, there is a relatively low probability that the control

channels will collide. If they are synchronised it may be necessary to time shift.

Carrier offsetting

When the HeNB downlink bandwidth is smaller than the overlay network, the HeNB

frequency could be frequency offset by several resource blocks in order to avoid

interference amongst control channels. There is still the risk of interference between data

and control channels, but this is less likely than control channels since the data channels will

not always be occupied. Low-power operators could agree to reduce power or mute

resource blocks likely to cause control channel interference.

Frequency partitioning

The PDCCH is a control channel which spans the entire bandwidth and is used for

interference estimation. Carrier offsetting will not avoid this, but it would be possible to use

only a portion of the band for this estimation. By only transmitting the PDCCH in these bands

and choosing different subbands for use by different HeNBs interference can be mitigated.

Other control signalling channels (PDCCH, PHICH, PCFICH, P-SCH, S-SCH, PBCH) could also be

RS

RP

Small

Indoor

loss

HeNB HUE

RS

RP

Large

Indoor loss

HeNBHUE

HeNB and HUE is located at edge and center of a

house respectively. In this case, interference from

HeNB to MUE will increase.

HeNB and HUE is located at center and edge of a

house respectively. In this case, an indoor coverage

will decrease.

MUE MUE

Internal wall

RSRP of a MeNB

RS

RP

Small

Indoor

loss

HeNB HUE

RS

RP

Large

Indoor loss

HeNBHUE






will decrease.

MUE MUE

Internal wall

RSRP of a MeNB

RS

RP

Small

Indoor

loss

HeNB HUE

RS

RP

Large

Indoor loss

HeNBHUE






will decrease.

MUE MUE

Internal wall

RSRP of a MeNB

RS

RP

Small

Indoor

loss

HeNB HUE

RS

RP

Large

Indoor loss

HeNBHUE






will decrease.

MUE MUE

Internal wall

RSRP of a MeNB


allocated to different frequency resources amongst neighbouring HeNBs to minimise

interference. Full implementation of this approach may depend on the use of Release 10 UEs

which are capable of aggregating frequency resources.

Uplink

Data Channels

Power control

To avoid creating uplink interference, HeNB-connected UEs (HUEs) should have their power set

based on path loss from the HUE to its nearest neighbour macro eNode-B. The path loss can be

estimated directly by the HUE from measurements of the macrocell Reference Signal Received

Power and by decoding messages which provide the macrocell transmitted power. The maximum

HUE transmit power can be set as a simple function of the estimated macrocell path loss or based on

a combination of the macrocell path loss and the path loss to the serving HeNB to balance any loss in

performance for both systems appropriately.

Control Channels

Time Offsets

As in the downlink, control channel interference can be mitigated by using time offsets to

the relevant control channels (PUCCH).

Additionally, the report discusses the concept of Hybrid Cells which are included in the 3GPP Release

9 specification. In a conventional Closed Subscriber Group cell, UEs which are not members of the

CSG gain no access to the cell and can suffer and create interference as a result, particularly when

they are at the edge of their serving cell (macro or HeNB) but close to the CSG HeNB. In hybrid

access mode, CSG UEs still have priority access to the CSG cell, but other UEs may be granted

admission on a conditional basis when the potential for interference is high. The result of a

simulation of interference between macrocells and hybrid cells is shown in Table*, where a

significant reduction in outage probability and increase in cell-edge throughput results relative to

conventional CSG HeNB even with adaptive transmit power.


Table 6: Improvement of performance using hybrid cells: results from simulation of interference between

HeNB and macro eNB with conventional CSG HeNB (source: 3GPP28

)

Outage Probability (SNR < -6 dB)

Worst 20% mobile throughput (kbps)

Median throughput (kbps)

No HeNB 12.7% 35 150

CSG HeNB with fixed Tx power of 8 dBm

18.9% 100 5600

CSG HeNB with adaptive Tx power

9.8 % 250 3300

Hybrid HeNB with fixed Tx power of 8

dBm

2% 900 5100

Hybrid HeNB with adaptive Tx power

3% 400 3400

This technique could be used in a slightly modified form as a form of ‘selective roaming’ amongst low-power concurrent operators, or even between low-power and high-power operators in an underlay configuration.

4.6 Stakeholder views

Although an extensive survey of stakeholders was not in scope for this study, we have made

contacted with a limited number of vendors of femtocells and other small cells in order to determine

their views on the feasibility of concurrent operation, and on the likely capabilities of practical

equipment for this band. Views on two particular issues were sought:

Appropriate transmit power levels to provide adequate coverage in a range of scenarios

The effectiveness of interference mitigation techniques in the specific scenario of concurrent operation in the same spectrum by different operators.

Only vendor stakeholders were consulted, namely:

Ubiquisys, a femtocell vendor

Picochip, a vendor of system-on-chips for femtocells

A tier one mobile infrastructure equipment vendor

The main points of the feedback received were as follows:

Ubiquisys

Much of the experience gained from deploying interference mitigation techniques in commercial 3G femtocell systems is also applicable, albeit in modified form, for LTE.

For home deployments, in the vast majority of homes, considerably less than +20dBm is required to provide good coverage. However, for public spaces and large enterprises considerably more may be required to provide adequate service, particularly if it is desired to achieve reasonable indoor coverage from outdoor cells.


In a concurrent operator scenario, notably in public places, some level of interference and associated performance degradation is inevitable if no coordination is applied, at least for dense deployments and where traffic levels are high. This degradation is not avoided by reducing the maximum transmit power, since cells will simply need to be closer together to avoid interference, so it would be better to give operators the freedom to deploy up to relatively high power levels, perhaps up to +30 dBm.

The interference mitigation techniques used in LTE for single operator deployments should be applicable to concurrent access, although techniques which rely on technical interfaces between operators are less likely to be useful.

Picochip

Transmit power should be higher than 20 dBm to provide useful coverage for applications beyond the home. The 3GPP local area base station specification of +24 dBm would be a useful starting point, combined with some moderate antenna gain.

Several of the conventional techniques for interference mitigation within a single operator network are likely to be inapplicable or at least less favourable for the case of concurrent shared access, in particular those involving open/hybrid access (which corresponds to roaming in the concurrent case) or inter frequency handover.

However, new techniques are continuously being developed and it is likely that existing techniques could be modified or enhanced to suit this scenario. No major specification or equipment changes would be needed to suit this case: essentially the same ‘toolbox’ of techniques would be used, but some additional optimisation might be needed potentially together with some additional agreements on approaches and caps of power levels in particular situations to be agreed amongst the relevant operators.

Infrastructure Equipment Vendor

Felt that dedicated spectrum for concurrent low power access was the preferred situation, notably because it would be difficult to avoid the ‘visiting problem’ in this case without roaming arrangements amongst operators.

If an underlay approach was adopted, however, it may be useful to offset the low power channel to overlap with two conventional channels, increasing the opportunity for frequency partitioning to avoid interference to the nearest macrocells in the overlay network.

In 3GPP simulations it is common to assume 30 dBm EIRP for outdoor femtocells and 20 dBm EIRP for indoor femtocells. For outdoor cells covering outdoor mobiles, 27 to 30 dBm EIRP is sufficient for good coverage along ‘street canyon’ environments.

However, providing good levels of coverage into buildings from outdoors is likely to require significantly higher powers, up to around 37 to 40 dBm EIRP. Such power levels are probably too high to justify a distinct low-power operation, and would certainly create excessive interference in an underlay situation, so around 30 dBm is probably a reasonable compromise.

The main interference mitigation technique of relevance is simply to apply downlink power control based on a variety of uplink and downlink measurements and an appropriate algorithm to relate the measurements to the maximum transmitted power. Other techniques of relevance include cell selection based on path loss, time offsets to avoid interference between control channels and the use of hybrid modes as specified in 3GPP TR 36.942.

http://www.3gpp.org/ftp/Specs/archive/36_series/36.942


There is scope for standardising advanced interference mitigation techniques to be suitable specifically for this concurrent operator case. It may be important to ensure that differing techniques are compatible with each other and do not lead to damaging instabilities, where for example all cells decide to increase their power towards the maximum to establish dominance over surrounding interferers.

In public spaces where multiple operators are deploying, coordination will be necessary to achieve good performance, although with some performance degradation uncoordinated deployment should be possible.

In the case of private environments such as homes, or offices with a single operator, the most challenging interference situation is when the closest adjacent home/office has a low power cell from another operator. Such situations need to be considered and appropriately handled whether there are 2 operators or 10 in the same spectrum, so the selection of the appropriate number of operators is mostly not related to technical issues.

4.7 We have made reasonable simplifying assumptions regarding

interference mitigation techniques based on the literature and

stakeholder feedback

It is not the prime purpose of the study to examine the detailed algorithms available for LTE

femtocells to mitigate and avoid interference. Nevertheless, our analysis of co-channel interference

needs to make reasonable assumptions concerning the potential performance of such techniques, at

least in the challenging conditions of main interest to this study.

We have therefore chosen to model two scheduling approaches , a dynamic scheduler and a random

scheduler. The operation of these is illustrated in Figure 4 for the case of two cells adjacent to each

other.


Figure 4: Illustration of the operation of two resource scheduling techniques for modelling purposes

In the case of the dynamic scheduler, the two cells are assumed to work cooperatively to minimise

the overlap of resources. This may be achieved by central coordination, by conventions concerning

which cells have priority over which resources, or via one of the distributed techniques described in

section 4.5. If cell 1 transmits in resources occupying a proportion p1of the available bandwidth B1, ,

while cell 2 transmits in a proportion p2of the available bandwidth, then the two cells have exclusive

use of a bandwidth (1-p2)B in the case of cell 1 and (1-p1) B in the case of cell 2. These resources

then suffer degradation only due to noise rather than interference. If (p1 + p2 ) < 1, there is no

interference between the cells. Otherwise, a bandwidth (p1 + p2 -1) Bis contended, and those

resources have degraded performance due to interference.

The overall throughput for cell 1 is therefore given by:

𝑇𝑝𝑢𝑡𝑑𝑦𝑛𝑎𝑚𝑖𝑐 = 𝑆𝐸 𝑆𝑁𝑅 × 1 − 𝑝2 × 𝐵 + 𝑆𝐸(𝑆𝐼𝑁𝑅) × 𝑝1 + 𝑝2 − 1 × 𝐵 ; 𝑝1 + 𝑝2 > 1

𝑆𝐸 𝑆𝑁𝑅 × 1 − 𝑝2 × 𝐵 ; otherwise

1 In practice some of the bandwidth is occupied by control channels. This overhead is accounted for in our

calculations but not shown here for simplicity.

Cell 1

Cell 2

System Bandwidth, B10 or 20 MHz

OccupiedResources, p1 B

OccupiedResources, p2 B

Contended Resources

= 0 if p1 + p2 < 1; = (p1 + p2 -1) B otherwise,

Total OccupiedResources

p1 B

Cell 1

Cell 2

System Bandwidth, B10 or 20 MHz

Total OccupiedResources

p2 B

Contended Resources,

average p1 p2 B

Dynamic Scheduler Random Scheduler

(1- p2)B

(1- p1)B


where 𝑆𝐸 ∙ is the spectrum efficiency as a function of SINR or SNR characteristic of the LTE system

in the relevant propagation conditions following the assumptions in section 4.1. A similar expression

applies for cell 2, although in our calculation we have assumed the same target throughput for both

cells and hence the same loading. SINR and SNR can be calculated from the relevant propagation

and equipment characteristics for any given pair of wanted and interfering paths. When computing

the separation between a wanted and interfering cell to achieve a given target throughput, our

calculations have varied the loading level p = p1 = p2to minimise the required separation distance.

In the case of the random scheduler, the resources are allocated by each cell without attempting to

completely avoid interference. Each cell may schedule within any of the resources, and if we assume

that the outcomes of those scheduling processes are statistically independent, then a mean

bandwidth of p1 p2 B is contended. The remaining (p1 B - p1 p2 B) resources (for cell 1) are

uncontended on average.

The overall throughput for cell 1 is therefore given by:

𝑇𝑝𝑢𝑡𝑟𝑎𝑛𝑑𝑜𝑚 = 𝑆𝐸 𝑆𝑁𝑅 × 𝑝1 − 𝑝1𝑝2 × 𝐵 + 𝑆𝐸(𝑆𝐼𝑁𝑅) × 𝑝1𝑝2 × 𝐵

Again the loading levels are adjusted to maximise the separation distance at which a given target

throughput can be achieved.

Note that the random scheduler can actually achieve higher performance than the dynamic

scheduler at high loading levels, because the interference is spread over the full system bandwidth,

reducing the average interference per resource, whereas the dynamic scheduler operation means

that some resource blocks suffer continuous interference. Where this is the case we have assumed

that the dynamic scheduler reverts to the performance of the random scheduler.


5 Power levels in line with 3GPP recommendations will ensure

good coverage for low-power shared access points

5.1 Study questions

Ofcom posed the study questions shown in Figure 5-1 to understand the technical conditions within

licenses for low power shared access channels. It was Ofcom’s working assumption that these

licenses would be no greater than the current 3GPP UTRA specification limits for Home BS. However,

the E-UTRA LTE specifications are still under development and the power limits for home base

stations are not yet defined. Therefore as part of this study, chapter 5 investigates the types of

power levels required to provide sufficient coverage for a number of given scenarios.

Figure 5-1 Ofcom study questions addressed in Chapter 5

Study questions

1.13 What minimum power level would be needed in order to provide

coverage in the following example scenarios:

• Indoor office environment

• Indoor public area

• Residential (home femtocell)

• Campus / business park (including use of external base station

antennas)

1.14 What depth of in-building coverage would be provided by an outdoor

base station operating at 20dBm e.i.r.p. in the campus scenario, assuming a

mast height of 5m or below?


Study approach

The study approach is outlined in the diagram in Figure 5-2 which defines the inputs to the model

and cell area output achieved according to the variable inputs. In the case of coverage analysis the

main input variables include the transmit power and different propagation models applied to the

different environments. The output cell area is then defined for a given fixed coverage confidence

which for this analysis shall be 78% at the cell edge, which is equivalent to 90% across the cell area

given simplifying assumptions concerning the cell geometry and shadowing variability.

Figure 5-2 Approach to low-power shared access coverage modelling

The expected output plots will be in the form shown in Figure 5-3 with the cell area or coverage

range as the output for a given transmit EIRP in dBm. The plots will be derived for a given SNR to be

achieved and thus range of user throughputs. This will help determine the types of throughput that

can be achieved for a given power level within the different types of propagation environment.

Figure 5-3 Output plots for coverage scenario analysis

Model

Propagation environment • Indoor to indoor within office environment • Indoor to indoor within public area • Indoor to indoor within a home• Short range outdoor and outdoor and outdoor to indoor for outdoor access point on a business park

Cell area

Variable transmit power

Fixed coverage confidence at 90%

Cell area

Power/dBm

Single user cell edge throughput/Mbps

10

41

91


5.2 Indoor office coverage – Study question 1.13

5.2.1 Scenario description and assumptions

The indoor office environment is intended to represent a low level open plan or partitioned office

that is considered typical for the UK. The diagram in Figure 5-4 represents an indoor office and

shows the paths from the low-power access point with the relevant propagation environment. The

indoor office scenario can come in many different forms, however in general an office layout will be

either ‘open plan’ or ‘partitioned’. In each case there will be the usual office furniture and

obstructions including desks, chairs, glass walls, plaster/brick/stud walls etc. The coverage level of an

indoor office environment must take into account the losses of the various obstructions. Our

example shows a low rise office building with a mix of open plan and partitioning and a low-power

access point on each floor which means the coverage is calculated through the walls.

The office floor area assumed for this scenario is a medium size low rise office of 840m2 capable of

accommodating 60 staff.

The transmit power from the 3GPP specification35 for a Local area BS max transmit power is 27dBm

and the antenna gain is based on a better unit being used in enterprise environments than that used

in low cost residential deployments.

A 10MHz channel is assumed the worst case for PSD out of our choice of 10MHz or 20MHz as the

maximum total transmit power is independent of bandwidth in the 3GPP specifications. The eNB

height of 2.5m assumes the unit is mounted on the ground floor on a wall or ceiling.

The propagation model used is the ITU-R P.1238 as this has been developed specifically for

modelling office environments and doesnot differentiate between open plan and partitioned offices.

A distance power loss coefficient is required for the analysis and is based on the ITU-R P.1238 2GHz

value.

The floor loss of 20.2 dB is the same as the residential and shopping centre scenarios for consistency

and shadowing is 10dB based on the value used in 3GPP R4-092042 when walls arenot explicitly

modelled.


Figure 5-4 Indoor office coverage area with enterprise eNodeB as low-power device

5.2.2 Results and conclusions

The results shown in Figure 5-5 are the coverage area results for downstairs and show a rapidly

increasing coverage area for the lower throughputs (10/20 Mbps). At moderate EIRP levels (E.g 10

dBm) the coverage area is 4000m2 to achieve 20 Mbps. A comparable coverage area for higher

throughputs (41/91 Mbps) requires higher EIRP levels in the order of 21 – 24 dBm in order to

maintain the target SNR. There is a wide gap between extent of coverage areas for higher and lower

throughputs for a given EIRP, for example at 10 dBm the difference in coverage area is almost

3000m2, which considerable in such an important commercial environment.

Figure 5-5 Indoor office downstairs coverage area

0

500

1000

1500

2000

2500

3000

3500

4000

0 10 20 30

Co

vera

ge a

rea

(m

^2

)

EIRP (dBm)

Downstairs Coverage Area

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


The results shown in Figure 5-6 are the coverage area results for upstairs. There is a clear distinction

between the upstairs and downstairs coverage area for an indoor office which is a requirement for

higher EIRP levels to achieve equivalent coverage for any given throughput. The higher throughput

levels do not exceed 250m2 even for the highest EIRP which restricts upstairs throughputs to the

lower values. Even at maximum EIRP the coverage area achieved is 1700m2 for a 20 Mbps

throughput which is less than half the downstairs coverage area.

Figure 5-6 Indoor office upstairs coverage area

The results shown in Figure 5-7 are the total coverage area results for both upstairs and downstairs

of the indoor office. The pattern closely matches the downstairs coverage area, however the extent

of total coverage is tempered by the reduced coverage upstairs. Our analysis suggests that the

maximum EIRP is not necessary for coverage of a single floor in a typical indoor office environment,

EIRP’s in the range of 17 – 20 dBm is sufficient to cover 1000-1500m2 for the higher throughputs

and well over 4000m2 for the lower throughputs.

0

500

1000

1500

2000

2500

3000

3500

4000

0 10 20 30

Co

vera

ge a

rea

(m

^2

)

EIRP (dBm)

Upstairs Coverage Area

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


Figure 5-7 Indoor office total coverage area

In Figure 5-8 the coverage range for downstairs shows a mix of gradual and steep increases in range

against increased EIRP. The pattern for the lower throughputs, up to 10 Mbps for example, show a

steeper increase in range for a given EIRP compared to higher throughputs from 20 Mbps upwards

which are more gradual. The range of coverage at maximum EIRP peaks at 160m for a 10 Mbps

throughput and reduces to 35m for 91 Mbps throughput. However the coverage range exceeds

200m at an EIRP limit of 23 dBm for the lower throughputs.

Figure 5-8 Indoor office downstairs coverage range

0

500

1000

1500

2000

2500

3000

3500

4000

0 10 20 30

Co

vera

ge a

rea

(m

^2

)

EIRP (dBm)

Total Coverage Area

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)

020406080

100120140160180200

0 5 10 15 20 25 30

Ra

ng

e (

m)

EIRP (dBm)

Downstairs range

1.5Mbps

2.6Mbps

4.5Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


In Figure 5-9 the coverage range for upstairs shows a similar pattern to the downstairs coverage

range but with the range limit reduced by a factor of about 4. This is due to the loss through the

ceiling and the walls upstairs. The coverage range for higher throughputs is no greater than 10m for

the maximum EIRP but it starts to improve for lower throughputs, with a 40m range for a 10 Mbps

throughput. Therefore, in the upstairs range case higher EIRP would be necessary to achieve the

higher throughput values but the extent these throughputs cover will be limited.

Figure 5-9 Indoor office upstairs coverage range

Conclusions

The following conclusions have been drawn based on the analysis of coverage area and coverage

range for a medium sized low rise office building for the particular scenarios covered in this chapter.

• Setting a maximum EIRP of 27dBm on the low power shared access channel (i.e. In line with the

3GPP Local area Base Station specification) would not limit operators in terms of achieving

maximum data rates at the range needed to cover a single floor medium sized office.

• This assumes office femtocells will be higher cost enterprise units and have an antenna gain of

3dBi

05

101520253035404550

0 10 20 30

Ran

ge (

m)

EIRP (dBm)

Upstairs range

1.5Mbps

2.6Mbps

4.5Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


• Coverage is more challenging on a second floor but it is assumed that most operators would

deploy a femtocell per floor in an office environment to provide adequate capacity for the

number of users per floor.

• Backing off the maximum EIRP to 18dBm would still provide adequate single floor coverage for

our example medium sized office of 840 m² at maximum data rates.

• It is assumed that for large office areas the operator would deploy multiple access points.

• It is worth noting that most 3G enterprise femtocells solutions support 16 or 32 voiceusers.

While the same limitations do not necessarily apply to LTE access points, as it is a packet

switched rather than circuit switched system, it is likely that most operators would deploy more

than one access point for our example medium sized office of 840 m² and 60 people for capacity

rather than coverage reasons. Operators may also deploy multiple access points to ensure

coverage in areas with larger losses than typical.


5.3 Residential coverage – Study question 1.13


The residential coverage scenario assumes a typical residential UK house in two particular types:

i) New build house ii) Old style house

The specific differences between the two house types are the wall material and the number of walls.

The wall type typically found within a new build house is a mixture of wood and plasterboard and

the wall type within an older style house will be brick and plasterboard. The two types of material

offer different wall penetration losses which could potentially affect the range of coverage area from

a home eNodeB.

Parameter Value Units

Local eNB tx power 0 to 20 dBm

Antenna gain 0 dBi

Bandwidth 10 /20 MHz

Home eNB height 0.5 m

Propagation model COST 231 multi wall model

rw,Distance between walls 4 for new builds

6 for old houses

m

Lw, Loss per internal wall 4.2 for new builds (Wood

and plaster)

7 for older houses

(Concrete /brick)

dB

Number of floors 2

Loss per floor 20. dB


Shadowing 4 dB

The average UK new build in 2009 had a room size of 15.9 m2with an average wall spacing of

approximately 4m. The value used in our analysis is in the region of 4.85m calculated from R4-

09204233 (assuming that the path loss equation used there is for 1800MHz and light walls). For

older houses we have assumed a larger wall spacing of 6m.

Typically the transmit power will be low for homeeNodeB’s i.e. 7dBm for ipAccess34product, the

3GPP Home eNodeB specification35 goes up to 20dBm. The maximum transmit power for local areas

base station is 24dBm but not likely to be used in this scenario.

The antenna gain is based on low cost unit using built in antenna and the eNB height of 0.5m which

assumes the unit is located downstairs on a desk.

A 10MHz channel is assumed the worst case for PSD out of the choice of 10MHz or 20MHz as the

maximum total transmit power is independent of bandwidth in the 3GPP specifications.

Table 5-1 Parameters used for residential coverage analysis

The following assumptions are made for the propagation environment:

Using COST 231 multi wall model in line with 3GPP simulation settings for LTE femtocells given in R4-092042.

Wall loss of 4.2 dB is calculated from the COST 231 wall loss for light walls at 900MHz and 1800MHz extended to 2.6GHz.

Floor loss of 20.2 dB is calculated from the COST 231 wall loss for light walls at 900MHz and 1800MHz extended to 2.6GHz.

Shadowing is the same value as used in 3GPP R4-092042.

The diagram in Figure 5-10 shows the transmit paths from a single home eNodeB for a residential

area which are analysed in this study and include:

Downstairs range and coverage area

Upstairs range and coverage area


Figure 5-10 Residential coverage scenario with indoor Home eNodeB as low power device


The following results have been produced using the path loss equation given below to establish the

power levels required for sufficient indoor residential coverage through a mix of wall and floor types

found in both new build style and older style houses and to achieve a 90% coverage confidence of

the cell area.

Equation 1Propagation environment - Indoor to indoor - COST 231 multi wall

Where LF = free space loss = -27.6 + 20 log( r3D)+ 20 log f (MHz)

Replace the middle term with wall loss as follows:

o Lw x r2D / rw

𝐿𝜏 = 𝐿𝐹 + 𝐿𝐶 + 𝐿𝑤𝑖

𝑊

𝑖=1

𝑛𝑤𝑖 + 𝐿𝐹𝑛𝑓 (𝑛𝑓+2)/(𝑛𝑓+1) −𝑏)


Residential femtocell range results10MHz vs. 20MHz in a new build house

The results shown in Figure 5-11 are the coverage range results for upstairs and follow a linear

pattern of increasing range with EIRP. The plot shows that for a 10 MHz channel the range is

marginally better than for a 20 MHz channel in all throughput example cases. However, it is also

noted that to achieve the higher throughputs of 20 and 40 Mbps the range decreases by 10m. The

maximum range achieved in our analysis is just over 35m at maximum EIRP for 10 MHz channel with

5 Mbps achievable throughput.

Figure 5-11 Residential coverage downstairs range in a new build house

In Figure 5-12, the coverage range for upstairs has a more gradual effect compared to the

downstairs range. This is due to the eNB being located downstairs and providing coverage upstairs.

The results show a lower maximum coverage range compared to downstairs for 10 and 20 MHz

channels (lower throughputs) of up to approximately 20m at maximum EIRP. For higher throughputs

(20/40 Mbps) at maximum EIRP, the maximum range achievable is around 15m.

0

5

10

15

20

25

30

35

40

0 5 10 15 20 25

Ran

ge (

m)

EIRP (dBm)

Downstairs range

5Mbps (10MHz)

10Mbps (20MHz)

20Mbps (10MHz)

40Mbps (20MHz)


Figure 5-12 Residential coverage upstairs range in a new build house

0

5

10

15

20

25

0 5 10 15 20 25

Ran

ge (

m)

EIRP (dBm)

Upstairs range

5Mbps (10MHz)

10Mbps (20MHz)

20Mbps (10MHz)

40Mbps (20MHz)


Residential femtocell coverage area resultsin a new build house

The results presented below are given for the coverage area of a new build house. It is assumed the

average floor area of a UK new build house is 76m2. However, the floor area of larger houses can be

up to 320m2 but we use the average floor area for our analysis.

The results are given for a 10 MHz channel except for 91 Mbps which is only achieved in a 20 MHz

channel. Figure 5-13 shows the results for downstairs coverage area which is based on achieving the

SNR at the cell edge required for a single user achieving the range of data rates shown. It can be

seen that for a relatively small EIRP (7 dBm) the coverage can be as much as 1800m2. As the EIRP

increases the coverage area increases with the lower throughput (10/20 Mbps) achieving over

2000m2 at 10 and 15 dBm EIRP. The resulting coverage area for higher throughputs (41/91 Mbps)

are lower by a difference of 1000m2 compared to the lower throughputs due to the higher SNR

required for increased throughput.

Figure 5-13 Residential downstairs coverage area in a new build house

Figure 5-14 shows the results for upstairs coverage area which is based on achieving the SNR at the

cell edge required for a single user achieving the range of data rates shown. It can be seen that the

coverage area upstairs is greatly reduced compared to downstairs (assuming eNB on ground floor)

due to the penetration through the floor and upstairs walls. At maximum EIRP the maximum

coverage area achieved for the higher throughputs is limited to 200m2. The coverage area is

improved, however for the lower throughputs up to a maximum coverage area if 1100m2 (10 Mbps).

0200400600800

100012001400160018002000

0 5 10 15 20 25

Co

vera

ge a

rea

(m

^2

)

EIRP (dBm)


10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


Figure 5-14 Residential upstairs coverage area in a new build house

The total coverage area is given in Figure 5-15 which assumes coverage of the entire house upstairs

and downstairs. The plot in Figure 5-13 is closely matched to Figure 5-15 with the pattern for the

lower throughputs achieving a higher total coverage area compared to the higher throughputs. It

can be suggested that for a single home eNodeB that providing coverage an average house it is not

necessary to transmit at maximum EIRP. An EIRP of 5-7 dBm is sufficient to cover 600m2 (almost

twice the average UK floor area of a larger house) and achieve throughputs up 91 Mbps.

Figure 5-15 Residential total coverage area in a new build house

0200400600800

100012001400160018002000

0 5 10 15 20 25

Co

vera

ge a

rea

(m

^2

)

EIRP (dBm)


10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)

0200400600800

100012001400160018002000

0 5 10 15 20 25

Co

vera

ge a

rea

(m

^2

)

EIRP (dBm)

Total Coverage Area

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


Residential femtocell range results – New build

The results in this section are for the coverage range for a home eNodeB upstairs and downstairs for

a new build house. The plots in Figure 5-16 show the coverage range for downstairs which increases

as the EIRP increases for a range of different throughputs. The pattern of increasing throughputs

reduces with range where the highest throughput (at maximum EIRP) achieves a maximum range of

17m compared to 41m for a 1.5 Mbps throughput at maximum EIRP.

Figure 5-16 Residential femtocell downstairs range (based on floor areas) – New Build

The pattern for the upstairs range is similar to the downstairs range except the maximum range for

each achievable throughput rate is reduced due to the penetration through the floor and the walls.

For example, the maximum range is 26m for 1.5 Mbps throughput at maximum EIRP compared to

41m for downstairs. This suggests even at modest EIRP levels (10-15 dBm) 10 – 20 Mbps can be

achieved upstairs at a range of 12- 15m.

0

5

10

15

20

25

30

35

40

45

0 5 10 15 20 25

Ra

ng

e (

m)

EIRP (dBm)

Downstairs range

1.5Mbps

2.6Mbps

4.5Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


Figure 5-17 Residential femtocell upstairs range (based on floor areas) – New Build

Residential femtocell coverage area results – Old houses

The results presented below are given for the coverage area of an old style house. It is assumed the

average floor area of a UK old style house is the same as a new build house at 76m2. In addition, we

assume the floor area of larger houses can be up to 320m2 but we use the average floor area for our

analysis.

Figure 5-18 shows the results for downstairs coverage area for an old style house which is based on

achieving the SNR at the cell edge required for a single user achieving the range of data rates shown.

It can be seen that for a relatively small EIRP (7 dBm) the coverage can be as much as 1500m2 to

achieve a 10 Mbps throughput. As the EIRP increases the coverage area increases with the lower

throughput (10/20 Mbps) achieving over 1800m2 at 10 and 15 dBm EIRP. The resulting coverage area

for higher throughputs (41/91 Mbps) are lower by a difference of 1000m2 compared to the lower

throughputs due to the higher SNR required for increased throughput.

0

5

10

15

20

25

30

0 5 10 15 20 25

Ran

ge (

m)

EIRP (dBm)

Upstairs range

1.5Mbps

2.6Mbps

4.5Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


Figure 5-18 Residential downstairs coverage area - Old house

It can be seen in Figure 5-19 the results for upstairs coverage area is greatly reduced compared to

downstairs (assuming eNB on ground floor) due to the penetration through the floor and upstairs

walls. At maximum EIRP the maximum coverage area achieved for the higher throughputs is limited

to below 200m2. The coverage area is improved for the lower throughputs up to a maximum

coverage area if 900m2 for a throughput of 10 Mbps.

Figure 5-19 Residential upstairs coverage area - Old house

0200400600800

100012001400160018002000

0 5 10 15 20 25

Co

vera

ge a

rea

(m

^2

)

EIRP (dBm)


10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)

0200400600800

100012001400160018002000

0 5 10 15 20 25

Co

vera

ge a

rea

(m

^2

)

EIRP (dBm)


10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


The total coverage area is given in Figure 5-20 which assumes coverage of the entire house upstairs

and downstairs. The plot in Figure 5-18 is closely matched to Figure 5-20 with the pattern for the

lower throughputs achieving a higher total coverage area compared to the higher throughputs. It

can be suggested that for a single home eNodeB that providing coverage within an average old style

house it is not necessary to transmit at maximum EIRP. An EIRP of 6-8 dBm is sufficient to cover 400-

500m2 and achieve throughputs up 91 Mbps downstairs with slightly lower throughputs upstairs.

Figure 5-20 Residential total coverage area – Old house

Residential femtocell range results – Old houses

The results in this section are for the coverage range for a home eNodeB upstairs and downstairs for

an old style house. The plots in Figure 5-21 show the coverage range for downstairs which increases

linearly as the EIRP increases for a range of different throughputs. The pattern of increasing

throughputs reduces with range where the highest throughput (at maximum EIRP) achieves a

maximum range of 16m compared to 37m for a 1.5 Mbps throughput at maximum EIRP.

0200400600800

100012001400160018002000

0 5 10 15 20 25

Co

vera

ge a

rea

(m

^2

)

EIRP (dBm)

Total Coverage Area

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


Figure 5-21 Residential downstairs coverage range – Old house

The pattern for the upstairs range is similar to the downstairs range except the maximum range for

each achievable throughput rate is reduced due to the penetration through the floor and the walls.

For example, the maximum range is 24m for 1.5 Mbps throughput at maximum EIRP compared to

37m for downstairs. This suggests even at modest EIRP levels (10-15 dBm) 10 – 20 Mbps can be

achieved upstairs at a range of 7- 15m.

Figure 5-22 Residential upstairs coverage range – Old house

0

5

10

15

20

25

30

35

40

0 5 10 15 20 25

Ra

ng

e (

m)

EIRP (dBm)

Downstairs range

1.5Mbps

2.6Mbps

4.5Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)

0

5

10

15

20

25

30

0 5 10 15 20 25

Ran

ge (

m)

EIRP (dBm)

Upstairs range

1.5Mbps

2.6Mbps

4.5Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


Conclusions

The following conclusions have been drawn based on the analysis of coverage area and coverage

range for the various residential scenarios covered in this chapter.


3GPP HeNB specification) would not limit operators in terms of achieving maximum data rates at

the range needed for even large new build houses in the UK.

• This assumes residential femtocells will be low costunits and have an antenna gain of 0dBi

• Coverage will be more challenging in older buildings compared to new builds. At EIRP 20dBm a

range of 16.5m at a max data rate of 91Mbps could be achieved downstairs and 6.66m upstairs.

This should be sufficient downstairs for most houses but may start to limit coverage at higher

data rates upstairs in some larger properties.

• Therefore the residential coverage scenario is not the limiting case for setting the maximum

transmit power and EIRP for the low power shared access channel, at least where interference

levels are negligible. In practice operators are likely to need to allow some interference margin

so greater power headroom may be required.

• Results show that the maximum transmit power level could be backed off significantly and still

achieve maximum data rates at the range needed for typical residential scenarios

• If the maximum transmit power was backed off to 10dBm (as suggested in 3GPP if activity in an

adjacent channel is detected) the downstairs range in older houses at a max data rate of

91Mbps would be 11m which is considered adequate for most houses.

• However, upstairs the range at this data rate and EIRP would be reduced to 3.7m so it is likely

that data rates would need to be backed off to 10 or 20Mbps (depending on 10MHz or 20MHz

BW allocation) to maintain reasonable upstairs coverage in larger houses.

• It is worth noting that in practice residential femtocells are highly cost sensitive and so the

maximum transmit power tends to be driven by heat dissipation constraints in the product

design. Typical 3G femto products on the market today have a transmit power of 7dBm (source:

ipAccess34) for example. However, this may change over time to accommodate higher transmit

powers and so any regulatory limits on a shared access channel will need to allow for this.


5.4 Indoor public area coverage – Study question 1.13


The diagram in Figure 5-23 represents an indoor public area which in the example below shows a

shopping centre layout. This example provides the challenging radio propagation environment due

to the mix of separate units, corridors and open spaces. Additionally, the walls tend to be brick/steel

reinforced and can be a challenging from a coverage perspective. The requirements to provide

sufficient coverage for an indoor public area may mean numerous indoor low power eNodeB’s using

our example of many different units. However, fewer indoor low power eNodeB’s may be required

in an open plan indoor public area due to reduced obstructions and losses.

The transmit power is from the 3GPP specifications for a Local area BS which for maximum transmit

power is 27dBm. The antenna gain is based on a better unit being used in shopping centres than that

used in low cost residential deployments and the eNB antenna height of 2.5m assumes the unit is

mounted on the ground floor on a wall or ceiling.

A 10MHz channel is the worst case for PSD out of our choice of 10MHz or 20MHz as the maximum

total transmit power is independent of bandwidth in the 3GPP specs.

The propagation model is the same as that used for the residential scenario but with a wider wall

spacing based on the average floor area of the 5 shop units currently advertised for rent in the

Sovereign Shopping Centre.

Wall loss of 4.2 dB is calculated from the COST 231 wall loss for light walls at 900MHz and 1800MHz

extended to 2.6GHz.

Floor loss of 20.2 dB is calculated from the COST 231 wall loss for light walls at 900MHz and

1800MHz extended to 2.6GHz.

Shadowing is the same value as used in 3GPP R4-092042.

The Sovereign Shopping Centre is approx 7800 m² on a single floor with 40 shops. As a comparison a

very large shopping centre such as the Trafford centre in Manchester (6th largest in UK) is 140,000 m²


Figure 5-23 Indoor public area coverage. Example shopping centre


The results shown in Figure 5-24 are the coverage area results for downstairs on a single level of the

shopping centre. There is a steady increase in coverage are for increasing EIRP up to the maximum

which can achieve over 50000m2 area for a 1.5 Mbps throughput. The coverage area for higher

throughputs (41/91 Mbps) reaches around 10000m2 at the maximum EIRP. In this case if we assume

an operator intends to achieve the maximum achievable throughput to ensure maximum capacity

then the maximum EIRP is required to provide the particular service and therefore more than one

access point is required.

Source: Sovereign Shopping Centre Weston Super Mare

05000

100001500020000250003000035000400004500050000

0 10 20 30

Co

vera

ge a

rea

(m

^2

)

EIRP (dBm)


1.5 Mbps

2.6 Mbps

4.5 Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


Figure 5-24 Indoor public area downstairs coverage area

The results shown in Figure 5-25 are the coverage area results for upstairs on a single level of the

shopping centre. There is a reduction in the coverage area from the low-power access point on the

ground floor as the maximum coverage area is 9250m2 for a 10 Mbps throughput at maximum EIRP

compared to 32500m2 coverage area for same service and EIRP downstairs. In this case we can

assume an operator would deploy a low-power access point on each floor to ensure satisfactory

service is delivered to its subscribers. This is validated by the much lower coverage area provided

upstairs for the higher throughputs (20 Mbps and above) at the maximum EIRP levels.

Figure 5-25 Indoor public area upstairs coverage area

The results shown in Figure 5-26 are the total coverage area results for both upstairs and downstairs

of the shopping centre. The pattern closely matches the downstairs coverage area as this is the

dominant scenario, however the extent of total coverage is tempered slightly by the reduced

coverage upstairs. Our analysis suggests that the maximum EIRP are likely to be necessary for

coverage and capacity of a single floor for a shopping centre environment. EIRP’s in the range of 20 –

24 dBm are sufficient to cover 10000m2 for the higher throughputs and well over 50000m2 for the

lower throughputs.

0100020003000400050006000700080009000

10000

0 10 20 30

Co

vera

ge a

rea

(m

^2

)

EIRP (dBm)


1.5 Mbps

2.6 Mbps

4.5 Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


Figure 5-26 Indoor public area total coverage area

In Figure 5-8 the coverage range for downstairs shows a gradual increase in range against increased

EIRP. The coverage range at maximum EIRP peaks at 130m for a 1.5 Mbps throughput and reduces

to 52m for 91 Mbps throughput.

Figure 5-27 Indoor public area downstairs range

In Figure 5-28 the coverage range for upstairs shows a similar pattern to the downstairs coverage

range but with the range limit reduced by a factor of about 2. This is due to the loss through the

ceiling and the walls upstairs. The coverage range for higher throughputs is no greater than 15m for

05000

100001500020000250003000035000400004500050000

0 10 20 30

Co

vera

ge a

rea

(m

^2

)

EIRP (dBm)

Total Coverage Area

1.5 Mbps

2.6 Mbps

4.5 Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)

0

20

40

60

80

100

120

140

0 5 10 15 20 25 30

Ra

ng

e (

m)

EIRP (dBm)

Downstairs range

1.5Mbps

2.6Mbps

4.5Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


the maximum EIRP but starts to improve for lower throughputs, with a 70m range for a 1.5 Mbps

throughput. Therefore, in the upstairs range case higher EIRP’s (>20 dBm) would be necessary to

achieve even the lower throughput values as the distances within a shopping centre become critical

for the number of low-power access points required for sufficient coverage range.

Figure 5-28 Indoor public area upstairs range

Conclusions

The following conclusions have been drawn based on the analysis of the coverage area and coverage

range for the shopping centre (indoor public area) scenarios covered in this chapter.


3GPP Local area basestation transmit power specification) would not limit operators in terms of

achieving maximum data rates at the range needed to cover a single floor medium sized

shopping centre.

• This assumes public area femtocells will be higher cost enterprise units and have an antenna

gain of 3dBi

• Coverage is more challenging on a second floor but it is assumed that most operators would

deploy a femtocell per floor in a public area environment to provide adequate capacity for the

number of users per floor.

0

10

20

30

40

50

60

70

80

0 5 10 15 20 25

Ran

ge (

m)

EIRP (dBm)

Upstairs range

1.5Mbps

2.6Mbps

4.5Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


• Backing off the maximum EIRP to 18dBm would still provide adequate single floor coverage for

our example medium sized shopping centre if 2 access points were used to cover the example

shopping centre.

• It is assumed that for larger shopping centres and public areas the operator would deploy

multiple access points.

• It is worth noting that most 3G enterprise femtocells solutions support 16 or 32 voiceusers.

While the same limitations do not necessarily apply to LTE access points as it is a packet

switched rather than circuit switched system it is likely that most operators would deploy more

than one access point in a shopping centre for capacity rather than coverage reasons.

Operators may also deploy multiple access points to ensure coverage in areas with larger losses

than typical.

5.5 Business park/Campus environment – Study question 1.13


The diagram in Figure 5-29 represents a business park which shows an outdoor low power eNodeB

serving both outdoor and indoor users. The relevant propagation environment for this scenario can

take many different forms. For example, the number and height of the buildings can vary as can the

wall types for each of the buildings. However for the purposes of this study our example shows low

rise office buildings on a business park with an outdoor low power eNodeB in the car park serving

both indoor and outdoor users.

The transmit power is from the 3GPP specifications for a Local area BS which for maximum transmit

power is 27dBm. The antenna gain is based on a better unit being used in an outdoor environment

than that used in low cost residential deployments and the eNB antenna height of 5m assumes the

unit is mounted on a lamp post outdoors.


total transmit power is independent of bandwidth in the 3GPP specsifications.

The propagation model for outdoor users assumes there is a short range LOS path between the

access point and UE in a reasonably cluttered environment and so uses ITU-R P.1411 for LOS in


street canyons. We add Depth 2 building penetration loss to this for indoor users which is based on

Depth 2 suburban loss from the 2G liberalisation study.

We only model ground floor users as these will have the worst LOS path back to the access point at

5m.Shadowing is the same value as used in 3GPP R4-092042 for macrocells.

Cambridge Science Park accommodates more than 90 companies and has a land area of approx

615,000 m² which forms the basis for the assumption used for this scenario.

Figure 5-29 Business park environment scenario, low rise serving indoor and outdoor users from an outdoor

eNodeB.


Outdoor users

The results shown in Figure 5-30 are the coverage area results for the outdoor coverage area

serving outdoor users. It can be seen that a wide area is covered outdoors for moderately low

powers. For example, the coverage area is almost 120000m2 at an EIRP of 15 dBm for a 10 Mbps

throughput. At the higher throughputs (41/91 Mbps) with a maximum EIRP the coverage area can

reach 50000m2 to 80000m2 which is sufficient to cover 1/16 of the Cambridge Science park.


Figure 5-30 Business park outdoor coverage area – Outdoor users

The results shown in Figure 5-31 are the coverage range results for the outdoor business park

scenario. The outdoor range can reach 800m at maximum EIRP for a 1.5 Mbps throughput down to

150m for the 41/91 Mbps throughputs. This range gives the cell radius for the low-power access

point and for the purposes of this study assume a 200m radius is adequate would mean the

maximum EIRP is not necessary to provide extensive outdoor coverage for and achieve a 20 Mbps

throughput

Figure 5-31 Business park outdoor coverage range – Outdoor users

020000400006000080000

100000120000140000160000180000200000

0 10 20 30 40

Co

vera

ge a

rea

(m

^2

)

EIRP (dBm)

Outdoor Coverage Area

1.5 Mbps

2.6 Mbps

4.5 Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)

0

100

200

300

400

500

600

700

800

900

0 10 20 30 40

Ran

ge

(m

)

EIRP (dBm)

Outdoor range

1.5Mbps

2.6Mbps

4.5Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


The results shown in Figure 5-32 are the coverage area results for the indoor coverage area serving

indoor users. It can be seen that where good indoor coverage can be expected the area covered is

reduced for a given EIRP, this is due to the building penetration losses assumed through the walls.

For example, the coverage area is reduced to 10000m2 for a 41 Mbps throughput at the maximum

EIRP compared to 75000m2 for the outdoor coverage area validating the building penetration losses.

Figure 5-32 Business park indoor user coverage area

The results shown in Figure 5-33 are the coverage range results for the indoor business park

scenario. The indoor range reaches 370m at maximum EIRP for a 1.5 Mbps throughput down to 40m

for the 41/91 Mbps throughputs. The coverage range offered to indoor users is greatly reduced

compared to the outdoor range, as expected and therefore higher EIRP levels are required to match

the coverage range of the outdoor cell but only for a 10 Mbps throughput or less and not a 20 Mbps

throughput or more.

020000400006000080000

100000120000140000160000180000200000

0 10 20 30 40

Co

vera

ge a

rea

wh

ere

go

od

ind

oo

r p

en

etr

atio

n c

an b

e e

xpe

cte

d (

m^

2)

EIRP (dBm)

Indoor user coverage area

1.5 Mbps

2.6 Mbps

4.5 Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


Figure 5-33 Business park indoor user coverage range

5.6 Depth of in-building coverage – Study question 1.14

5.6.1 Approach to depth of in-building coverage

The approach to analysis of depth of in-building coverage is captured in Figure 5-34 which differs

slightly from Figure 5-2 ,the overall approach to coverage modelling. The modelling of depth of in-

building coverage uses two particular scenarios to capture the effects of different angles of arrival to

the wall of the building.

A. The campus scenario is used to derive the direct angle of arrival assuming 90° B. The street scenario is used to derive the oblique angle of arrival assuming 45°

0

50

100

150

200

250

300

350

400

0 10 20 30 40

Ran

ge f

rom

low

po

we

r ac

cess

po

int

to

bu

ildin

g fo

r go

od

in

do

or

pe

ne

trat

ion

(m

)

EIRP (dBm)

Indoor user range

1.5Mbps

2.6Mbps

4.5Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)

Fixed transmit power

Fixed coverage confidence at 90%

Model

Propagation environment scenarios:• Campus scenario (direct angle of arrival)• Street (oblique angle of arrival)

Depth of in-building coverage


Figure 5-34 Approach to depth of in building coverage analysis

The expected output plots will be in the form shown in Figure 5-35 with the depth of coverage in

metres as the output for a given distance between the building and access point. The plots will be

derived for a given SNR to be achieved and thus range of user throughputs. This will help determine

the types of throughput that can be achieved for a given depth of coverage at two defined power

levels.

Figure 5-35 Expected results output

The diagrams in Figure 5-36 and Figure 5-37 show both example scenarios for depth of in-building

coverage. The scenarios represent the two main differential angles of the radio signal when

penetrating into a building from an outdoor low power eNodeB. Scenario A shows the direct angle of

arrival from the outdoor low-power eNodeB to the indoor users with little or no angle deviation

from the base station and therefore the users are receiving the most direct signal possible. In

contrast scenario B shows the oblique angle of arrival from the outdoor low-power eNodeB to the

indoor users and the received signal arrives at a given angle to the users from eNodeB, this is due to

the close proximity of the base station to the building resulting in potentially reduced in building

coverage as shown in the diagram.

The EIRP is based on two values as requested by Ofcom, 20dBm and 29dBm based on the 3GPP35

Local area base station maximum transmit power of 24dBm and an antenna gain of 5dBi based on a

better unit being used in outdoor environments than indoors. The antenna height of 5m assumes

the access point is mounted on a lamp post outdoors.

Depth (m)

Distance between cell and building

10

41

91

Single user cell edge throughput/Mbps



total transmit power is independent of bandwidth in the 3GPP specs.

The propagation model is the same as used for the Business park coverage scenarios but with COST

231 building penetration loss added to give greater accuracy of building penetration at varying

distances between the buildings.

Building penetration loss based on COST 231 LOS building penetration loss and assumes:

External wall losses in line with COST 231 values for a concrete wall with a normal size window

Internal wall made of wood and plaster and keeping the same value as used in the residential and shopping centre scenarios

Wall separation based on an average office space available in IQ science park in Cambridge split into 14 rooms to include open plan areas, meeting rooms, labs etc.

We only model ground floor users as these will have the worst LOS path back to the access point at 5m.

Shadowing is the same value as used in 3GPP R4-092042 for macrocells.


EIRP 20, 29 dBm

Bandwidth 10 MHz

Propagation model ITU-R P.1411 and COST 231

Mobile height 0.5 m

Basestation height 5 m

External wall loss 7 at 90 °

20 at 0°

dB

Internal wall loss 4.2 dB

Internal wall separation 8 m

Loss per floor Ground floor only

Shadowing 8 dB

Table 5-2 Parameters for depth of in-building coverage scenarios


Scenario A – Direct angle of arrival

Figure 5-36 Depth of in-building coverage business park with direct angle of arrival 90° due to typical

distance between building and access point

Scenario B – Oblique angle of arrival

Figure 5-37 Depth of in-building coverage with oblique angle of arrival 45° due to close distance between

buildings and access point

Direct angle

Direct angle

Oblique angleOblique angle



Direct angle of arrival

The results shown in Figure 5-38 are the in-building depth for a direct angle of arrival at 20 dBm

EIRP. The pattern shows decreasing depth of in-building coverage for increasing perpendicular

distance between the building and access point. The depth of in-building coverage also decreases

with increasing throughput. For example, at 20m distance between the building and access point the

depth achieved is 10m for a 91 Mbps throughput compared to a 70m depth at the same distance for

1.5 Mbps throughput. Lower throughputs can achieve 20-40m depth of in-building coverage even at

distances up to 100m but the closer to the access point is to the building the better depth of in-

building coverage is achieved for all throughputs.

Figure 5-38 Depth of in-building coverage 20 dBm Direct angle of arrival

The results shown in Figure 5-39 are the in-building depth for a direct angle of arrival at 29 dBm

EIRP. The difference compared to the 20 dBm scenario shows greater depth of in-building coverage

for comparable distances between the building and access point. Using our same example, at 20m

distance between the building and access point, the depth achieved is 27m for a 91 Mbps

throughput compared to a 90m depth at the same distance for 1.5 Mbps throughput.

0

20

40

60

80

100

120

0 20 40 60 80 100 120

In b

uild

ing

cove

rage

de

pth

(m)

Perpendicular distance between building and access point (m)

In building depth - EIRP 20dBm, direct angle of arrival

1.5 Mbps

2.6 Mbps

4.5 Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


Figure 5-39 Depth of in-building coverage 29 dBm Direct angle of arrival

Oblique 45 degree angle of arrival

The results shown in Figure 5-40 are the in-building depth for an oblique angle of arrival at 20 dBm

EIRP. The pattern shows decreasing depth of in-building coverage for increasing perpendicular

distance between the building and access point. The depth of in-building coverage also decreases

with increasing throughput. Using the same example as the direct angle, at 20m distance between

the building and access point there is no depth recorded for a 91 Mbps throughput. Using 41 Mbps

at 20m distance between the building and access point achieves a 7m depth of in-building coverage

compared to a 60m depth at the same distance for 1.5 Mbps throughput.

This can be compared to the depth achieved for the same values of the direct angle of arrival which

was 70m for 1.5 Mbps compared to 60m for Mbps at 20m distance for the oblique angle of arrival.

Therefore at an oblique angle of arrival the depth of coverage is reduced compared to the direct

angle due to attenuation of the diffracted path of the signal from the low-power access point.

0

20

40

60

80

100

120

0 20 40 60 80 100 120

In b

uild

ing

cove

rage

de

pth

(m)


In building depth - EIRP 29dBm, direct angle of arrival

1.5 Mbps

2.6 Mbps

4.5 Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


Figure 5-40 Depth of in-building coverage 20 dBm oblique 45 degree angle of arrival

The results shown in Figure 5-41 are of the in-building coverage depth for an oblique angle of arrival

at 29 dBm EIRP. The pattern compared to the 20 dBm scenario shows some increase in the coverage

depth inside the building. Using our same example, at 20m distance between the building and access

point the depth achieved is 19m for a 91 Mbps throughput compared to a 79m depth at the same

distance for 1.5 Mbps throughput.

Figure 5-41 Depth of in-building coverage 29 dBm oblique 45 degree angle of arrival

0

10

20

30

40

50

60

70

80

90

100

0 20 40 60 80 100 120

In b

uild

ing

cove

rage

de

pth

(m)


In building depth - EIRP 20dBm, oblique 45 degree angle of arrival

1.5 Mbps

2.6 Mbps

4.5 Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)

0

20

40

60

80

100

120

0 20 40 60 80 100 120

In b

uild

ing

cove

rage

de

pth

(m)


In building depth - EIRP 29dBm, oblique 45 degree angle of arrival

1.5 Mbps

2.6 Mbps

4.5 Mbps

10 Mbps

20Mbps

41Mbps

91 Mbps (20MHz)


Conclusions


3GPP Local area basestation transmit power specification) would not limit operators in terms of

achieving maximum data rates to outdoor users at typical microcell ranges (i.e. 100m)

• This assumes outdoor low power access points will be higher cost microcell like units and have

an antenna gain of 5dBi

• Coverage is more challenging for indoor users. At an EIRP of 29dBm the target cell edge data

rate would need to be backed off to 20Mbps to achieve the target microcell range of 100m.

• Results for indoor penetration support the above point that getting good coverage indoors even

at an EIRP of 29dBm is challenging.

5.7 Summary

Study question Answer

1.13 What minimum power level would be needed in order to provide coverage in the following example scenarios:

• Indoor office environment

• Indoor public area • Residential (home

femtocell) • Campus / business

park (including use of external base station antennas)

Indoor office environment: An EIRP of 27dBm l (i.e. In line with 3GPP Local area basestation specification and a 3dBi antenna gain) easily provides maximum data rates at the range to cover a single floor medium sized office and could be backed off to 18dBm. Indoor public area: An EIRP of 27dBm l (i.e. In line with 3GPP Local area basestation specification and a 3dBi antenna gain) would provides maximum data rates at the range to cover a single floor medium sized shopping centre (8,000 m²). In practice operators are likely to deploy more than one access point for capacity reasons in the office and public areas and so the transmit power levels here could potentially be backed off even more. Residential (home femtocell): An EIRP of 20dBm provides a range of 16.5m at a max data rate of 91Mbps downstairs and 6.66m upstairs. This is sufficient downstairs for most houses but may start to limit coverage at higher data rates upstairs in some larger properties. Campus / business park (outdoor access point): Outdoor users – AnEIRP of 29dBm on the low power shared access channel (i.e. In line with the 3GPP Local area basestation transmit power specification) would provide maximum data rates to outdoor users at typical microcell ranges (i.e. 100m) Indoor users - An EIRP of 29dBm would require the target cell edge data


rate to be backed off to 20Mbps (in 10MHz) to achieve the target microcell range of 100m and good indoor penetration. Coverage is limited by the indoor penetration of outdoor basestations. Recommend the maximum EIRP is set at 29dBm in line with this.

1.14 What depth of in-building coverage would be provided by an outdoor base station operating at 20dBm e.i.r.p. in the campus scenario, assuming a mast height of 5m or below?

Both 20dBm EIRP and 29dBm EIRP were examined as a maximum transmit power of 24dBm and an antenna gain of 5dBi for outdoor local area basestations are discussed in 3GPP and were also highlighted by stakeholders. 29dBm is required to achieve maximum data rates into the building at ranges from the building of more than 20m. For direct angles of arrival, like a campus scenario, at 29dBm EIRP and a distance of 50m from the building an in building depth of 16m is achieved for peak data rates in 10MHz and 10m for peak data rates in 20MHz. At 100m from the building this is reduced negligible levels of 1m. For an oblique angle of arrival, like a street scenario , the indoor penetration is worse. At a 50m perpendicular distance from the building and 45 degree angle of arrival the in building depth for maximum data rates at 10MHz is 6.5m. At 20MHz the range from the building needs to be reduced to 45m to get just 2m of indoor coverage


6 Co channel interference is manageable via appropriate antenna

heights,dynamic scheduling and compromises in target

throughputs

6.1 Study questions

Ofcom posed the study questions shown in Figure 5-1 to understand the impact of co-channel

interference on devices in the proposed low power shared access channel.

Figure 6-1 Ofcom study questions addressed in Chapter 6

The following example scenarios shown in Table 6-1 have been analysed to answer these study

questions.

Study questions

1.17 What is the minimum separation distance between buildings where low-power

networks can be deployed without operator coordination becoming necessary.

1.18 What limits should placed on the maximum height of outdoor antennas, for the

purpose of limiting interference to other low-power networks (our working assumption is

5m)?

1.19 In the case of underlay low-power networks, what restrictions on antenna placement

would be needed in order to minimise interference to the co-channel wide-area network, e.g.

minimum distance from macrocell antennas, restriction to indoor placement?

1.20 Based on an assumption that some low-power operators will deploy outdoor antennas

on low-power networks, what is the minimum distance before the frequency (or resource

blocks) can be re-used by indoor networks? What is the minimum separation distance for

deployment of co-channel low-power indoor and outdoor networks without operator

coordination becoming necessary.


Study

Question

Deployment scenario Aggressor Victim Requirement

1.17 Minimum separation

distance between low

power networks in shared

spectrum

Single femto

eNodeB

(interfering)

Single femto UE

in adjacent

building (DL)

Separation distance

in metres between

buildings against

throughput

degradation.

Single high power

femto UE

(interfering)

Single femto

eNodeB in

adjacent building

(UL)

1.18 Maximum height of

outdoor antennas

Single outdoor

femto eNode B with

variable antenna

height

Single outdoor

femto UE on

another low

power outdoor

network (DL)

with 5m antenna

Antenna height in

metres against

separation distance

between low power

networks

1.19

Restrictions on (indoor)

antenna placement for

underlay approach due to

interference to co-

channel wide area

network

Single indoor femto

UE on limit of

reception

(interfering)

Single outdoor

macro eNodeB

(UL)

Separation distance

between aggressor

and victim against

throughput

degradation.

Single indoor

eNodeB

Single outdoor

UE (DL)

1.20

Minimum distance from

outdoor low power

antennas before

frequency (or RB’s) can be

re-used by indoor

networks

Single outdoor

eNodeB

Single indoor UE

(DL) served by

indoor Low-

Power eNodeB

Separation distance

between aggressor

and victim against

throughput

degradation.


Single outdoor UE Single indoor

eNodeB (UL)

Table 6-1: Co channel interference scenarios to address study questions

Study approach

The study approach is outlined in the diagram in Figure 5-2. For each of the study questions posed

we have defined an example scenario with an aggressor, the device causing the interference, and a

victim, the device suffering interference. The following inputs are required to the model:

Transmit power levels for the aggressor based on our earlier coverage results for target throughput in the scenario

SNR at the victim receiver based on the target cell edge throughput in the absence of interference

The allowable SINR at the victim receiver based on the allowable degraded cell edge throughput in the presence of interference

Device parameters such as antenna gains at both the aggressor and victim and the victim receiver noise figure

Channel conditions including the most appropriate propagation model for the scenario being modelled and values for the shadowing standard deviation

Figure 6-2 Approach to low-power shared access co-channel modelling

As LTE devices will only suffer interference on the resource blocks that overlap with the aggressor

device it would be pessimistic to calculate the target throughput based on achieving the target SINR

across all resource blocks. Instead we have modelled the two schedulers; a random scheduler and a

dynamic scheduler, as described in section 4.2.1. Based on the two schedulers we estimate the

number of contended and uncontended resource blocks available, determine how much of the

target throughput would be carried on the uncontended resource blocks at the SNR and then

Scenario mix

Scenarios:• Indoor/Outdoor• Small cell/Macro cell• Dedicated / underlay / hybrid• 10MHz / 20MHz• Uplink or downlink

Data rate degradation (based on SINR degradation)

Aggressor

Victim

2.6 GHzequipment parameters


calculate a target SINR for the remaining contended resource blocks to carry the remainder of the

target throughput. A link budget between the aggressor and victim is then used to determine the

separation distance necessary to achieve this SINR.

The target throughputs are based on the throughputs that would be achieved by a single cell edge

user at 78% confidence to match the criteria applied in our coverage analysis. Our analysis reflects

very much a worst case scenario of users on the cell edge in the most challenging conditions. The

actual degradation across the entire cell will be depend of the resulting SINR distribution across the

cell which was beyond the scope of this study.

The model output illustrates the expected degradation in throughput for different target

throughputs at varying separation distances as shown in Figure 5-3.

Figure 6-3 Output plots for coverage scenario analysis

6.2 Building separation – Study question 1.17


Figure 6-4 illustrates the example scenario modelled to estimate the minimum building separation

required to minimise interference between low power networks. We have selected a residential

scenario of two neighbours both using low power access points. In the downlink scenario the UE is

at the edge of coverage of its own low power network and suffers interference from the

neighbouring access point. On the uplink the access point in the second house is desensitised by

uplink interference from the UE in the first house which results in cell shrinkage in the second house.

% Reduction in throughput

Separation distance (m)

Single user cell edge original target throughput

40Mbps

20Mbps

10Mbos


Figure 6-4: Example scenario used for estimating building separation to minimise interference between low power

networks

We have modelled the downlink scenario to answer this study question as:

The path loss and transmit power levels involved are the same in both cases

Target throughputs will be more challenging on the downlink

The victim receiver parameters are worse on the downlink scenario

In our model we have made the following assumptions:

Power control is applied on the downlink so that the aggressor transmit power varies in line with our coverage results for a 10m upstairs range. This range assumes a worst case that the victim UE and access point are on either side of a typical 4 bedroom detached house.

A 10MHz bandwidth is modelled as this will have the maximum PSD

Path loss is modelled using ITU P.1411 street canyon model

Two external wall losses of 7dB each are applied. This represents a concrete external wall with a window.

A 3dB fade margin is applied to the target SINR based on 4dB shadowing standard deviation at both the victim and aggressor.

A 0dBi antenna gain is applied at both the aggressor and victim. This assumes residential access points will be low end devices.

Interference from operator

2 on the DL

Femto UE on limit

of reception

Wanted

UE close to adjacent eNodeB

Operator 1

Operator 2

Cell shrinkage due to uplink

noise rise from nearby femto UE of operator 1

Operator 2 Femto UE on

limitof reception

Wanted


Operator 1

Operator 2

Downlink

Uplink


The target SINR is set to receive the allowable degraded throughput at a 78% confidence limit for cell edge users. This is in line with the confidence limit used in our coverage analysis.


Figure 6-5 shows the results for the separation distance that would be required between the two

neighbouring houses in our example scenario to achieve different allowable levels in throughput

degradation (shown on the y axis). The allowable throughput degradation is a proportion with

respect to the original target throughput for a single cell edge user at the victim receiver in the

absence of interference. Each throughput (shown by line colour) represents a different SNR starting

assumption at the victim receiver. We assume that the aggressor is also targeting this rate and sets

its transmit power to the higher levels required as the target throughput increases in the absence of

noise increases.

Figure 6-5: Separation distances required between two houses each containing a low power access point to

achieve the target throughput degradation shown

The dashed lines on the graph show the results using the random scheduler. The solid lines show

the performance of the dynamic scheduler which detects interference and schedules in un-

contended resource blocks as much as possible to reach the target throughput. When the target

degraded throughput allows for capacity to be shared equally amongst the contending networks, a

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

0 10 20 30 40 50 60 70

Allo

wab

le d

egr

adat

ion

in

th

rou

ghp

ut

wrt

no

ise

-lim

ite

d

sce

nar

io

(%)


10 Mbps Dynamic scheduler20 Mbps Dynamic scheduler41 Mbps Dynamic scheduler10 Mbps Random scheduler20 Mbps Random Scheduler41 Mbps Random scheduler


zero separation distance can be achieved as both the aggressor and victim can use 50% of resource

blocks un-contended to achieve the target throughput. This is better than a random scheduling

approach which in this scenario must have an allowable degradation of 70% before the separation

distance can be removed and the required degraded throughput is achieved on average by random

scheduling. As the allowable percentage degradation increases and more contended resources

need to be used to achieve the degraded throughput, the dynamic scheduler reverts to the random

approach as this achieves better performance than dynamically detecting interference and sharing

capacity equally. Therefore the dashed and solid lines converge on the graph.

If an allowable degradation in throughput of 50% is not acceptable to operators for this scenario

then a separation distance between low power networks of greater than 25m is required. Our

results indicate that degradations of less than 20% are not achievable in the presence of

interference at the data rates examined. Generally this separation distance decreases as the original

target throughput at the victim receiver decreases as they will require lower target SINRs. However,

there is a minimum limit on the target SINR of -2dBm to recover the PDCCH. When this limit is

reached the separation distances for the higher original throughputs with higher victim SNRs start to

approach those of lower target throughputs with lower SNRs.

Conclusions

The following conclusions can be drawn from the results for this scenario:

With the use of a dynamic scheduler that identifies interference and targets un-contended resource blocks the data rate at the cell edge will be degraded by the number of interfering access points i.e. in this case 2 networks were modelled so the throughput can be expected to be halved.

If the above degradations in throughput are acceptable no minimum separation distance is required between buildings provided this smart interference mitigation technique is applied.

If the above degradations are not acceptable then the minimum separation distance is likely to be upwards of 25m depending on the signal level at the victim receiver. This will be difficult to enforce in practice.

It should be noted that this scenario represents a worst case scenario of interference to a cell edge user from a nearby access point. The SINR across the victim cell will vary with distance, propagation environment conditions and wanted signal strength. These results only represent the degradation to a cell edge user and does not represent the degradation in throughput across the entire cell.


6.3 Outdoor antenna height – study question 1.18


Figure 6-6 shows the example scenario analysed to determine the maximum recommended mast

height for outdoor lower power access point masts. This represents two adjacent streets with two

different low power operators providing service into each of the houses along the street. The

aggressor is an outdoor access point whose downlink signal provides interference to a UE using the

network in the adjacent street. In this scenario we have only modelled the impact of interference on

an outdoor user. For indoor users the affect would be similar as although the interference signal is

reduced by the building loss the wanted signal at the indoor victim UE would also be degraded by

the same amount.

Figure 6-6: Example scenario analysed for determined maximum antenna heights for outdoor low power

access points

In our model we have made the following assumptions:

Power control is applied on the downlink so that the aggressor transmit power varies in line with our coverage results for a 50m range between the outdoor access point and building and achieving reasonable (depth 2) building penetration. This range assumes that a series of access points are regularly spaced on lamp posts to cover a street.


Path loss is modelled using the ITU P.1411 non line of sight Walfisch Ikagami model using a 9m roof top height and 14m street width.

A 8dB fade margin is applied to the target SINR based on an 8dB shadowing standard deviation at both the victim and aggressor.

A 5dBi antenna gain is applied at the aggressor and assumes outdoor access points will be more high end than indoor equivalents. A 0dBi antenna gain is assumed at the UE.

Street A Short distance with diffraction over

rooftops

5m

UE’s on limit of coverage and at different heights

Outdoor femto

eNodeB at up to 15m

Outdoor femtoeNodeB at

standard 5m



Figure 6-7 shows how the required separation distance between the outdoor low power access

point aggressor and the victim UE varies with the mast height of the aggressor if a 50% degradation

in throughput at the victim UE is allowed. As in the previous scenario the dynamic scheduler can

provide the degraded throughput with no minimum separation distance if the system capacity is

divided amongst the number of contending networks i.e. 50% in this scenario of two networks. As

this is an over roof top scenario there is an assumption that there will always be at least a building

and half a road between the victim and aggressor and so a separation distance of less than 19m

cannot be reached.

Figure 6-8: Separation distances required between an outdoor access point at different mast heights and a

victim low power UE to achieve a to achieve a 50% degradation on the target throughput shown

In the case of the random scheduler we can see that separation distances and hence interference

rises sharply beyond a 10m mast height. This is commensurate with the 9m building height used in

the scenario.

Conclusions

The following conclusions can be drawn from the results for this scenario:

0.00

2.00

4.00

6.00

8.00

10.00

12.00

14.00

16.00

0 20 40 60 80 100 120 140 160 180

He

igh

t o

f ag

gre

sso

r o

utd

oo

r ac

cess

po

int

(m)

Separation distance required for 50% allowable throughput degradation (m)

10 Mbps Dynamic scheduler20 Mbps Dynamic scheduler41 Mbps Dynamic scheduler10 Mbps Random scheduler

20 Mbps Random Scheduler



o As in the residential case, if a throughput degradation proportional to the number of interfering access points is acceptable then the outdoor antenna height does not impact separation distance as a dynamic intelligent scheduler can be used to avoid interference.

o The minimum separation distance of 19m is the minimum achievable in this scenario as this is a non line of sight over roof top scenario. This distance represents a building plus ½ a road.

o In the example scenario, the separation distance required increases steeply for heights beyond 10m which is commensurate with the rooftop height of 9m used in this scenario.

o Limits on outdoor antenna heights should be set in line with the height of surrounding buildings. The residential scenario will be the limiting case so we recommend a height of 10-12m.

6.4 Interference to macrocells in an underlay approach – study question 1.19


Figure 6-9 shows the example scenario analysed to understand uplink interference to macrocell

basestations from low power networks in an underlay spectrum approach where spectrum is shared

between the two networks. In this scenario the UE of the low power network is at the edge of

coverage and transmitting at maximum power. This desensitises a nearby macrocell basestation and

causes cell shrinkage. The impact of this interference will degrade the throughput of users across

the entire macrocell.

Figure 6-9: Uplink interference to a macrocell basestation from a low power UE in an underlay scenario

Distant User can’t access

network (dead zone)

Wanted Signal

InterferenceCell

shrinkage due to uplink

noise rise

Large distance and/or significant obstruction

Short distance and/or good

LOS

Restrictions on power of indoor mobiles due to location to outdoor macro antenna

High power UE

Outdoor macro eNodeB


In our model we have made the following assumptions for this scenario:

The UE transmits at its maximum power of 23dBm. We have also allowed for the 3GPP interference mitigation approach where transmit power is dropped to 10dBm if the low power network detects a a macrocell that it is sharing spectrum with.


Path loss is modelled using the ITU P.1411 with a single external wall loss of 8dB representing a concrete external wall with a window

A 5.5dB fade margin is applied to the target SINR based on an 4dB and 8dB shadowing standard deviation at the aggressor and victim respectively.

A 0dBi antenna gain is applied at the aggressor. A 14dBi antenna gain is assumed at the victim.

Figure 6-10 shows the downlink interference experienced by a macrocell network in an underlay

spectrum approach. Here the low power access point is located close to a macrocell UE on the edge

of coverage. In the worst case the macrocell UE will be visitor to the house where the low power

access point is being used but isn’t able to roam onto the low power network.

Figure 6-10: Downlink interference to macrocell UE from a low power network


Power control is applied on the downlink so that the aggressor transmit power varies in line with our coverage results for a 10m upstairs range. This range assumes a worst case that the victim UE and access point are on either side of a typical 4 bedroom detached house.



A 5.5dB fade margin is applied to the target SINR based on an 4dB and 8dB shadowing standard deviation at the aggressor and victim respectively.

Wanted Signal

Interference

Wide area UE at edge

of cell

Large distance and/or significant obstruction


LOS

Restrictions on placement or power of low-power indoor antennas due to interference to co-channel wide area UE

Outdoor macro eNodeB


A 0dBi antenna gain is applied at the aggressor and victim.


Uplink results

Figure 6-11 and Figure 6-12 show the resulting separation distances that would be required between the

victim macrocell basestation and aggressor low power network UE to achieve a range of allowable

throughput degradations. It can be seen that the dynamic scheduler provides lower separation distances

than the random scheduler, as was the case in earlier scenarios, and that a zero separation distance can be

achieved if operators are willing to share capacity in proportion to the number of contending networks.

Figure 6-11 assumes that the aggressor UE is at the edge of coverage and transmitting at a maximum power

level of 23dBm. Figure 6-12 shows the separation distances achieved if the UE transmit power is reduced to

10dBm. This is to illustrate the impact of a power cap on low power network devices if they detect that they

are in the region of a macrocell that they are sharing spectrum with. This is an approach that has been

suggested in 3GPP and is applied in the transmit power standards for HeNBs when other wide area networks

are detected in adjacent channels.

Figure 6-11: Separation distances required between victim macrocell basestation and low power network UE

aggressor to achieve the target throughput degradation shown for a 23dBm aggressor power

0%

10%

20%

30%

40%

50%

60%

70%

0 500 1000 1500 2000 2500 3000 3500 4000 4500

Allo

wab

le d

egr

adat

ion

in

th

rou

ghp

ut

wrt

no

ise

-lim

ite

d

sce

nar

io

(%)


4.5 Mbps Dynamic scheduler

10 Mbps Dynamic scheduler


4.5 Mbps Random scheduler


16 Mbps Random scheduler


Figure 6-12: Separation distances required between victim macrocell basestation and low power network UE

aggressor to achieve the target throughput degradation shown for a 10dBm aggressor power

If throughput degradations of less than 50% are required in these scenarios then separation

distances in the range of 500m to 2km and 1km – 4km will need to be applied for the case with and

without the 10dBm transmit power cap respectively. It is unlikely that distances in this region will be

practical to enforce.

Downlink results

Figure 6-13shows the resulting separation distances that would be required between the victim

macrocell UE and aggressor low power network access point to achieve a range of allowable

throughput degradations. Compared to the uplink results examined previously these separation

distances are much smaller due to the reduced antenna gain and height of the victim receiver. The

uplink interference results should therefore drive choices on the separation distances between low

power access networks and macrocell networks.

0%

10%

20%

30%

40%

50%

60%

70%

0 500 1000 1500 2000 2500

Allo

wab

le d

egr

adat

ion

in

th

rou

ghp

ut

wrt

no

ise

-lim

ite

d

sce

nar

io

(%)









Figure 6-13: Separation distances required between victim macrocell UE and low power network access

point aggressor to achieve the target throughput degradation shown

Conclusions

The following conclusions can be drawn from the results for these scenarios:

• The separation distances required on the uplink interference scenario are much larger than on the

downlink. The impact of uplink interference is to desensitise the macrocell BS for all UEs and so the

impact is more significant to the entire cell than the downlink interference case.

• Separation distances in the range 500m-2km would be required if a 20%-40% throughput

degradation is acceptable in the example scenario. This relies on a 10dBm power cap being applied

to the aggressor.

• If a throughput degradation proportionate to the number of contending networks sharing capacity is

acceptable the interference can be managed by smart scheduling and a separation distance is not

required. For example in this case a throughput degradation of 50% would need to be acceptable.

• It is unlikely that a wide area operator would be willing to accept such a degradation in throughput

or that these separation distances would be practical. It is more likely that the low power access

0%

10%

20%

30%

40%

50%

60%

70%

0 20 40 60 80 100 120 140

Allo

wab

le d

egr

adat

ion

in

th

rou

ghp

ut

wrt

no

ise

-lim

ite

d

sce

nar

io

(%)









points would have to refrain from transmitting when in the area of a macrocell if an underlay

approach was applied.

• Other interference mitigation in the underlay scenario could include a power cap on low power

access points detecting a macrocell network or to allow the UE causing the downlink interference to

roam onto the macrocell network.

6.5 Separation distance between outdoor and indoor low power networks –

study question 1.20


Figure 6-14 shows the downlink interference experienced by an indoor low power network from an

outdoor low power access point. Here the low power network UE is located at the edge of coverage

with a weak signal and suffers interference from a nearby low power access point outdoors that it is

sharing spectrum with.

Figure 6-14: Example scenario for downlink interference from an outdoor low power network to an indoor

low power network


We have modelled two levels of transmit power at the aggressor . The first assumes a fixed transmit power of 24dBm based on teh 3GPP limit for local areas basestations. As an alternative we have also investigated the case when power control is applied at the aggressor to adjust transmit power levels to those required to reach target throughputs at a 50m range and with reasonable (depth 2) in building penetration as shown by our coverage results.


Interference

Minimum distance between outdoor and indoor

Limit of coverage

Outdoor femtoeNodeB

5m

Downlink case



A 5.5dB fade margin is applied to the target SINR based on an 4dB and 8dB shadowing standard deviation at the victim and aggressor respectively.

A 0dBi antenna gain is applied at the victim. A 5dBi antenna gain is assumed at the aggressor.

Figure 6-15 shows the example scenario analysed to understand uplink interference to an indoor

low power network from an outdoor low power network that it is sharing spectrum with. In this

scenario the UE of the outdoor low power network causes downlink interference to the access point

of the indoor network resulting in cell shrinkage.

Figure 6-15: Example scenario for uplink interference from an outdoor low power network to an indoor low

power network


We assume that the outdoor UE is at the edge of coverage and transmitting at a maximum power level.



A 5.5dB fade margin is applied to the target SINR based on an 4dB and 8dB shadowing standard deviation at the victim and aggressor respectively.

A 0dBi antenna gain is applied at the victim and aggressor. .


Downlink results

Figure 6-16 and Figure 6-17 show the separation distances required between an outdoor access

point for a low power network causing interference to an indoor UE on a low power network that it

Minimum distance between outdoor and indoor

Limit of coverage

Outdoor femtoUE

Uplink caseTolerable interference

Victim femtoeNodeB near

window


is sharing spectrum with. Figure 6-16 shows the result when a constant maximum transmit power of

24dBm, in line with the maximum specified by 3GPP for LTE local area basestations. Figure 6-17

shows the result when power control is applied at the aggressor and the transmit power is adjusted

to provide the target throughputs at 50m range with reasonable (depth 2) in building penetration in

line with our coverage results for outdoor access points.

Figure 6-16: Separation distances required between victim indoor UE and outdoor access point aggressor to

achieve the target throughput degradation shown a constant 24dBm transmit power is applied at the

aggressor

0%

10%

20%

30%

40%

50%

60%

70%

0 100 200 300 400 500 600 700 800 900 1000

Allo

wab

le d

egr

adat

ion

in

th

rou

ghp

ut

wrt

no

ise

-lim

ite

d

sce

nar

io

(%)









Figure 6-17: Separation distances required between victim indoor UE and outdoor access point aggressor to

achieve the target throughput degradation shown when power control is applied at the aggressor

Uplink results

Figure 6-18 shows the resulting separation distances for the equivalent uplink interference scenario

where the outdoor UE casese interference to the indoor access point. It can be seen that the

separation distances in this case are lower than those required for the downlink interference

scenario. Therefore the downlink interference results should be used to guide separation distances

between indoor and outdoor access points.

0%

10%

20%

30%

40%

50%

60%

70%

0 50 100 150 200 250 300 350 400 450 500

Allo

wab

le d

egr

adat

ion

in

th

rou

ghp

ut

wrt

no

ise

-lim

ite

d

sce

nar

io

(%)









Figure 6-18: Separation distances required between victim indoor access point and outdoor UE aggressor to

achieve the target throughput degradation shown

Conclusions

The following conclusions can be drawn from the results for these scenarios:

• The separation distances required in the downlink interference scenario is much larger than on the

uplink.

• Separation distances in the range 100m-450m are required depending on the target throughput and

acceptable degradation level. This relies on power control being applied at the aggressor.

• As in previous scenarios, if a degradation in throughput proportionate to the number of contending

networks is acceptable then no separation distance is required and the interference can be managed

via dynamic scheduling.

6.6 Summary

Study question Answer

0%

10%

20%

30%

40%

50%

60%

70%

0 50 100 150 200 250 300

Allo

wab

le d

egr

adat

ion

in

th

rou

ghp

ut

wrt

no

ise

-lim

ite

d

sce

nar

io

(%)









1.17 What is the minimum separation distance between buildings where low-power networks can be deployed without operator coordination becoming necessary.

For the scenario analysed of two low power access points in adjacent houses, the minimum separation distances will be upwards of 25m assuming data rates of at least 10Mbps are targeted in interference free conditions and a throughput degradation of less than 50% is required at the cell edge. This will increase for more interference sources. With the use of a dynamic scheduler that identifies interference and targets un-contended resource blocks a zero separation distance can be achieved but the data rate will be degraded in proportion to the number of contending access points. In the example of 2 access points no separation distance would be needed if a degradation in throughput of 50% is acceptable at the cell edge.

1.18 What limits should placed on the maximum height of outdoor antennas, for the purpose of limiting interference to other low-power networks (our working assumption is 5m)?

Interference increases significantly once the antenna height is beyond the height of the surrounding buildings. Assuming a limiting case of a residential scenario this would suggest a limit of 10-12m.

1.19 In the case of underlay low-power networks, what restrictions on antenna placement would be needed in order to minimise interference to the co-channel wide-area network, e.g. minimum distance from macrocell antennas, restriction to indoor placement?

A much larger distance is required to mitigate uplink interference to the macrocell basestation than to macrocell UEs. Also the impact of uplink interference is to desensitise the macrocell basestation for all UEs and so the impact is more significant to the entire cell than the downlink interference case. In the analysed scenario of a single UE from a low power network operating on the edge of coverage a separation distances in the range 500m-2km would be required if a 20%-40% throughput degradation is acceptable. As previously described a zero separation distance is feasible if smart scheduling and a degradation in edge of cell throughput in proportion to the number of networks contending for resources is acceptable.

1.20 Based on an assumption that some low-power operators will deploy outdoor antennas on low-power networks, what is the minimum distance before the frequency (or resource blocks) can be re-used by indoor networks? What is the minimum separation distance for deployment of co-channel low-power indoor and outdoor networks without operator coordination becoming necessary.

The scenario analysed looks at interference between a single outdoor low power network and single indoor low power network. The separation distances required to minimise downlink interference are much larger than those on the uplink. In the scenario analysed, separation distances in the range 100m-450m are required to minimise interference on the downlink depending on the target throughput and acceptable degradation level. As previously described a zero separation distance is feasible if smart scheduling and a degradation in edge of cell throughput in proportion to the number of networks contending for resources is acceptable.


7 Trade-offs relating to spectrum quantity

This chapter examines how the quantity of spectrum potentially made available to a low power

shared access band impacts the potential capacity of that band in terms of maximising the number

of operators and volume of traffic that could be supported. Our analysis so far has examined

coverage and interference in specific low power shared access scenarios. This chapter draws on

these results to illustrate how different sharing regimes and spectrum quantities will impact the

overall capacity of a shared access channel.

7.1 Study questions

The FDD portion of the 2.6GHz band contains 2 x 70MHz of paired spectrum. If part of this is made

available for a low power shared access channel the benefits of such a shared access channel need

to be balanced against the reduction in spectrum potentially available for wide area deployments.

A low power shared access channel could be assigned as a completely separate “designated”

channel from those used by wide area operators with macrocell networks. If 10MHz were allocated

to the low power shared access band this would allow three 2 x20MHz wide area licences to also be

made available allowing the maximum number of wide area operators to benefit from peak LTE data

rates. However, low power operators would be limited to the user experience offered at 10MHz.

Alternatively the shared channel could be introduced as an “underlay” which shares spectrum with

an existing macrocellular network either in a coordinated (such as via a geolocation database or

direct connection) or autonomous (such as sensing spectrum gaps) manner. This would allow the

shared access channel to be extended to 20MHz while preserving three 2 x 20MHz licences for wide

area use but with the consequence of one of the wide area networks having potentially reduced

capacity due to interference from the low power shared band.

A “hybrid” spectrum allocation which is made up of 10MHz of dedicated spectrum and 10MHz of

underlay spectrum is also feasible as a third alternative.

The options for a low power shared access band examined by this study include:

2 x 10MHz Vs. 2 x 20MHz

Underlay Vs. Hybrid Vs. Dedicated

In each option the SINR experienced both in the macrocell and femtocells will be degraded in

different ways which will have a knock-on effect on capacity within these cells. The choice of


bandwidth for the shared channel may also influence how readily resources can be shared amongst

different femtocell and macrocell users and again impact capacity.

Some example bandplans which could result from the introduction of the low power licences are shown in

Figure 7-1.

Figure 7-1: Example bandplans incorporating low-power concurrent allocation

Ofcom have posed the following questions in this area of de.termining the “Optimum quantity of

spectrum”:

Study questions

1.9 For both the designated spectrum approach and the underlay approach,

assess the capabilities of the spectrum to support several concurrent low-power

shared access operators if the quantity made available is:

a) 2 × 10 MHz

b) 2 × 20 MHz

1.10 In particular, the assessment should include:

• what traffic capacity could be supported;

• how many concurrent low-power operators could be accommodated (noting

Frequency (GHz):

2.50

0

2.51

0

2.52

0

2.53

0

2.54

0

2.55

0

2.56

0

2.57

0

2.58

0

2.59

0

2.60

0

2.61

0

2.62

0

2.63

0

2.64

0

2.65

0

2.66

0

2.67

0

2.68

0

2.69

0

2.70

0

2.71

0

2.72

0

2.73

0

2.74

0

2.75

0

Pass iveRadar above 2.7 GHz

Service

High power FDD licencees only

20 MHz

HP

20 MHz

HP

20 MHz

HP

20 MHz

LP

20 MHz

LP

Hybrid 20MHz low-power

20 MHz

HP

20 MHz

HP

20 MHz

HPTDD

20 MHz

HP

20 MHz

HP

20 MHz

LP

20 MHz

LP

20 MHz

HP

10

MHz

HP

20 MHz

HP

10

MHz

HPUnderlay 20MHz low-power

20 MHz

HP

20 MHz

HPTDD

20 MHz

HP

20 MHz

LP

10

MHz

HP

20 MHz

LP

TDD

Dedicated 20MHz low-power

20 MHz

HP

20 MHz

HP

20 MHz

HP

10

MHz

HP

20 MHz

HP

20 MHz

HP

10

MHz

LP

Dedicated 10MHz low-power

TDD20 MHz

HP

20 MHz

HP

20 MHz

HP

10

MHz

LP

20 MHz

HP

FDD UL TDD FDD DL

20 MHz

HP

20 MHz

HP

20 MHz

HP

10

MHz

HP

20 MHz

HP

20 MHz

HP

20 MHz

HP

10

MHz

HP

TDD


the working assumption of eight licensees); and

• what is the impact of the spectrum quantity on an operator’s frequency re-

use within a geographic area.

1.11 What techniques are available for licensees to manage the spectrum sharing

between low-power operators?

7.2 Factors impacting capacity in small cells

The capacity that can be supported by small cells differs to that expected in traditional wide area

networks in two key areas:

Improvements in capacity via improvements in the spectrum efficiency density achievable due to:

o A higher density of cells o An improved SINR distribution across the cell o Improvements in performance of technologies such as MIMO in indoor

environments where small cells are more prominent. o Fewer users per cell giving lower contention rates and better user experience

Reductions in capacity via interference due to the uncoordinated nature of small cells deployments which will vary with the:

o Number of operators sharing the low power channel o Spectrum approach taken with interference from macrocells being an issue in the

underlay and hybrid approaches.

The following two sections examine each of these areas.

7.2.1 Spectrum efficiency density in small cells

The capacity of a cellular network is a function of three key elements Figure 7-2:

The spectrum used to deliver the service

The technology which delivers bits over the air

The topology of the cells which comprise the network


Figure 7-2: Capacity depends on a combination of spectrum, technology and topology of the

network

Equation 2 shows how these three factors can be combined to represent capacity in a network. Here

spectrum efficiency is an indicator of the network technology’s ability to use the allocated spectrum

to deliver services which meet the required quality criteria.

Capacity density

= Quantity of spectrum

x Cell Spectrum Efficiency

x Cell density

[bits per second per km2]

[hertz] [bits per second per hertz per cell]

[cells per km2]

Spectrum Technology Topology Equation 2

The spectrum efficiency (i.e. throughput per Hertz of spectrum) achieved in practice in a cell

depends highly on distribution of the signal quality across the cell, commonly known as the SINR

distribution. Figure 7-3, Figure 7-4 and Table 7-1 below gives an example of the difference in

throughput and in turn spectrum efficiency due to the improved SINR distribution encountered in

smaller cells.


Figure 7-3 : SINR distributions for small and large cell topologies (Source: Ericsson[36

], Qualcomm[37

]

Figure 7-4: Analysing the benefit of higher peak rates: Comparison of SINR required for higher rates with the

probability of achieving that SINR in a given deployment scenario. Sources: Rysavvy research[38

],

Ericsson[39

], Qualcomm[40

]

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

-20 -10 0 10 20 30

Pro

bab

ility

th

at S

INR

< o

rdin

ate

SINR, dB

Large cell

Small Cell

0%

1%

2%

3%

4%

5%

6%

7%

8%

-30 -20 -10 0 10 20 30 40

Pro

bab

ility

of

SIN

R o

ccu

rin

g

SINR, dB

Large cell

Small Cell

0

1

2

3

4

5

-30 -20 -10 0 10 20 30 40

Lin

k Le

vel P

erf

orm

ance

, b

ps/

Hz/

use

r

SINR, dB

2x2

1x2

benefit of higher peak rate only available at higher SINRs

more likely to havehigher SINRs in small cell topologies


Table 7-1: Cell spectrum efficiency for normal and higher peak rate technologies, in large and small cell

environments (Source: Real Wireless analysis)

In addition to these improvements in capacity due to an improved SINR distribution, we would

expect technologies such as MIMO which rely on high amounts of multi path to operate better in

indoor environments where small cells are more likely to be used.

As there is a strong link between the cell spectrum efficiency and cell density, the spectrum

efficiency density is therefore frequently used to show the combined effect of technology and

topology between different cell sizes or deployment options.

Capacity density = Quantity of spectrum x Spectrum Efficiency

Density

[bits per second per km2]

[hertz] [bits per second per hertz per km2]

Spectrum Technology &Topology Equation 3

Figure 7-5 illustrates the potential capacity gains from small cells and is based on simulations in

three different test environments used in the ITU-R IMT-advanced evaluation. This shows that

smaller cell topologies have higher cell spectrum efficiencies resulting from the better SINR

distribution, lower terminal device speeds and channel rank. The difference in spectrum efficiency

density is even greater. It should be noted that the three options shown below are not comparable

in cost. More sites and equipment would be needed for the small cell topologies, so the higher

capacity of the microcell network comes at a higher cost.

It should be noted that the ITU indoor hotspot test environment is based on isolated cells at an

office or hotspot. This may be a simplified environment compared to a large scale deployment of

femtocells in an office block coexisting on the same carrier as outdoor macrocells.

These 3GPP simulation results show a x2.3 improvement in cell spectrum efficiency and x54

improvement in cell spectrum efficiency density between indoor hotspots and urban macrocells.

Cell Spectrum Efficiency

b/s/Hz/cell

Topology 1x2 2x2

Large Cell 0.73 0.78 6%

Small Cell 1.18 1.34 14%

Benefit 61% 73%

TechnologyBenefit

0.000.200.400.600.801.001.201.401.60

1x2 2x2

Technology

b/s

/Hz/

cell

Large Cell

Small Cell


Figure 7-5 - Impact of network topology and environment on spectrum efficiency for a 4x2 MU-MIMO

20MHz LTE-A system (Source: 3GPP[41

])

It is also worth noting that due to the improved SINR distribution in smaller cells that more users are

likely to benefit from the peak data rates and enhanced user experience offered by a 20MHz channel

than in a larger macrocell.

7.2.2 The impact of uncoordinated shared access on capacity

Wide area networks rely on the operator planning the network to coordinate interference and

performance across sites. This is partially automated by the X2 link in LTE which exchanges the

overload indicator, neighbouring cell tables etc. However, in small cells coordination between

access points or operators can’t be relied upon and it is not clear that an X2 link between the

potentially large volume of low power access points to automate coordination would be feasible.

Therefore more dynamic interference mitigation techniques are needed to minimise interference

and maximise performance and capacity.

Section 4.xx has already examined interference mitigation techniques in small cells. These are

mainly based around:

power control to ensure low power devices do not transmit any further than necessary to maintain range for their users

dynamic resource allocation to minimise collisions of resource blocks

As shown by our co channel interference assessment in chapter 6, if dynamic scheduler which

detects interference and targets un-contended resource blocks is applied in shared networks the

capacity in interference limited situations will be shared amongst the number of adjacent access

points causing interference. This approach relies on operators being willing to accept degradations

6.1

3.22.6

1957

27736

0

500

1000

1500

2000

2500

3000

0

1

2

3

4

5

6

7

Indoor hotspot Urban micro Urban macro

Spe

ctru

m e

ffic

ien

cy d

en

sity

b/s

/Hz/

km2

Ce

ll sp

ect

rum

eff

icie

ncy

b/s

/Hz/

cell

LTE-A Rel-10, FDD, DL, L=2, MU-MIMO 4x2, Antenna configuration C


in throughput in overlapping deployments proportional the number of contending networks and

being willing to back off target data rates in a fair way when interference is detected.

As not all deployments from all low power operators will overlap the number of low power licencees

does not place a direct cap on the capacity of small cells However, obviously the likelihood of

overlapping deployments will increase with the number licencees and so this should be kept to a

manageable level.

Assuming all operators did deploy in the same target area the best case scenario would be where

the operators all deploy access points at the same location and the capacity per operator would be

the capacity divided by the number of operators as shown below. However, without coordination of

access point placement between operators the capacity is likely to be degraded by a factor greater

than the number of operators due to interference.

Figure 7-6: Illustration of capacity shared amongst operators

In addition to interference from other low power access points, in the underlay or dedicated

spectrum scenarios there will be additional interference from a macro network which will further

reduce the capacity of the low power network and vice versa. Our co channel interference analysis

has shown that separation distances of 500m to 2km are required between macrocells and indoor

networks to achieve throughput degradations of less than 50% for the case of two contending

networks. This is assuming that a 10dBm transmit power cap is placed on low power networks

detecting macrocells in their area. These separation distances are likely to be difficult to enforce in

practice. In the full underlay scenario it is more likely that the macrocell would be given priority and

the low power access point would be expected to schedule resources to avoid the macrocell. The

Number of shared resource blocks for eight concurrent operators

82 3 4 5 6 71

10 MHz = 50 RB’s 8 operators = 6 RB’s per operator

6No of RB’s 6 6 6 6 6 6 6

82 3 4 5 6 71

8 operators = 12 RB’s per operator

12 12 12 12 12 12 12 12

No of RB’s

20 MHz = 100 RB’s


low power network therefore would not be guaranteed access to any spectrum in areas where

coverage from macrocell networks are already at reasonable levels.

The hybrid spectrum approach is an improvement on the underlay situation as it at least guarantees

access to half of the low power shared band and benefits from a boost in user experience in areas

where macrocell coverage is poor. The hybrid approach also opens the way for better interference

mitigation. One approach discussed in 3GPP involves scheduling the common channels for low

power devices in the dedicated portion of the hybrid spectrum allocation and making use of

spectrum aggregation features in LTE release 10 devices to schedule data when in the overlapping

portion of spectrum when no neighbouring wide area networks are detected. This avoids

overlapping control channels between the wide area and low power networks and so there is no

concern of the macrocell devices missing paging requests, call terminations etc..

A code of practice amongst operators making use of the shared band is could be used to ensure that

capacity is maximised. This could potentially include:

Implementing intelligent scheduling schemes that back off target throughputs and share capacity where neighbouring low power access points and, in the underlay case, macrocells are detected (similar to the dynamic scheduler assumed in our co channel interference assessment).

Implementing power control so that low power access points and associated UEs never transmit at higher powers than required for the target throughput.

Ensuring low power access points target low speed UEs so that the UE channel and it’s resource allocation isn’t changing frequently with time and can be tracked by other adjacent uncoordinated access points attempting to share capacity.

If separation distances are used to mitigate interference, the power spectral density (PSD) used on control channels should not exceed the PSD on the shared data channels. This is to ensure that the target SINR for control channels is achievable at the victim receiver if separation distances are set based on assumptions on the transmit power of the aggressor at a target throughput.

In the underlay or hybrid situation, priority should be given to the macrocell network with the low power access point reducing its power and accepting throughput degradations when a macrocell is detected.

Although the licencees should be able to choose the technology which they deploy, efficient

interference mitigation amongst operators are likely to require the use of the same time and

frequency resource block sizes amongst operators and compatibility with dynamic scheduling

approaches and environment measurements by both the UE and local area access points.

7.2.3 Summary of trade offs across spectrum options

Table 7-2 summarises the advantages and disadvantages of the different spectrum approaches for

the low power shared access channel taking into consideration the factors that impact capacity in


small cells, discussed in the previous two sections, and the impact on availability of spectrum for

wide area licences.

Spectrum option Pros Cons

2 x 10MHz - dedicated Allows three 2x20MHz wide

area licences to be made

available.

Restricts low power networks

to half the peak data rates of

wide area networks and less

than half the capacity.

Reduced opportunity to use

fractional frequency reuse

schemes to avoid interference.

2 x 20MHz - dedicated Improved user experience in

low power networks compared

to 10MHz

Larger bandwidth allows for

better resource scheduling and

less collisions on resource

blocks between cells.

Enhanced opportunity for

fractional reuse schemes to

improve isolation between

concurrent operators.

Reduces capacity available for

wide area operators

2 x20MHz – underlay Improved user experience in

low power networks compared

to 10MHz

Larger bandwidth allows for

better resource scheduling and

less collisions on resource

Having the underlay overlap

two high power networks

allows increased opportunity to

Extra interference from the

macrocell as well as adjacent

femtocells to consider

Capacity of third wide area

licence is degraded by more

than 40% in the case of two

contending networks if

separation distances of 500m -

2km are not maintained


avoid mutual interference

between high power and low

power networks

(assuming a 10dBm power cap

on low power networks in the

region of macrocells)

If low power network is a

secondary user of the band

they are not guaranteed access

in all locations.

2 x 20MHz - hybrid Low power networks

guaranteed access to at least

10MHz even when close to

macrocells.

Avoiding collisions of control

channels between macro and

low power devices is made

easier.

Capacity of third wide area

licence is degraded by more

than 40% in the case of two

contending networks if

separation distances of 500m -

2km are not maintained

(assuming a 10dBm power cap

on low power networks in the

region of macrocells)

Setting licence conditions to

ensure appropriate sharing

with the overlapping wide area

licencee will be challenging.

Table 7-2: Summary of advantages and disadvantages of different spectrum choices for a low power shared

access channel

During our recent study for Ofcom into 4G capacity we have seen that operators with 20MHz

allocations may benefit from trunking gains compared to those with 10MHz allocations if high user

throughputs are considered. This is an area with few results to date but is worth monitoring in

terms of the potential benefits of a 2x20MHz allocation over a 2x10MHz allocation for the low

power shared access channel.

7.3 Example scenarios of how capacity is impacted by spectrum choice

In this section we give examples of how the improvements and degradations in capacity anticipated

in a low power shared network and discussed earlier would manifest themselves in practice. Our

two example scenarios include:


A residential scenario

A shopping centre scenario

In each case we examine how capacity would be influenced by:

The number of operators sharing the spectrum

The spectrum approach choice of hybrid, dedicated or underlay

7.3.1 Residential scenario

Figure 7-7 shows a typical residential deployment for low power access networks. In this scenario it

is likely that each household will own one access point and that any contention for capacity will

come from SINR degradation due to interference from neighbouring households.

Figure 7-7: Example residential scenario for examining capacity reductions

As shown in our analysis of separation distances between low power access points in chapter 4, if an

intelligent scheduler is used capacity gets shared amongst the number of contending low power

networks. In this case capacity would therefore become a function of the number of houses

qualifying as “dominant interferers” rather than the number of operators or licencees of the shared

access channel.

Our co channel interference analysis also indicates that interference from outdoor low power

networks to indoor low power networks will be an issue if separation distances of upwards of 200m

are not maintained. In this example scenario it is feasible that a fourth operator is providing a

broadband service to the street via outdoor cells within this separation distance which would further

degrade the capacity of the indoor access points.

Wanted


Operator 1

Operator 2

FSPL

Operator 3


A third source of interference and capacity degradation would be from macrocell networks in the

underlay scenario as most houses are likely to be within the large separation distances of 500m –

2km indicated by our co channel interference results.

7.3.2 Shopping centre scenario

Figure 7-8 shows an example scenario of a shopping centre where a large number of operators may

potentially deploy contending low power networks. The best case scenario for overlapping

deployments is shown first. This is where the operators all deploy access points at the same location

and the capacity per operator would be the capacity divided by the number of operators. Below this

is the worst case scenario of without coordination of access point placement between operators

where access points are placed nearby to UEs at the edge of coverage. Our co channel analysis has

examined a similar situation to this worst case scenario in form of separation distances required

between buildings with residential access points. However, in this case the degradation in

performance is likely to be worse as there will be no external walls separating the victim receiver

and aggressor. However, our co channel interference results show that if a dynamic scheduler that

detects interference and targets un-contended resources in a fair way that accepts throughput

degradations in proportion to the number of contending networks, then separation distances aren’t

necessary. With this in mind, in our example shopping centre scenario, there would be an incentive

for operators to apply roaming and network sharing in overlapping areas rather having the expense

of rolling out their own access points without any additional benefits in capacity.


Figure 7-8: Example shopping centre scenario for multiple low power operator deployments

The same observations about degradations due to interference from outdoor low power networks

and any overlaid macrocell networks as were discussed for the residential scenario above will also

apply to this scenario.

7.4 Summary

Study Question Summary answer

1.9 For both the designated spectrum

approach and the underlay approach,

assess the capabilities of the spectrum to

support several concurrent low-power

shared access operators if the quantity

Smaller cells provide significant capacity

improvements due to:

• A higher density of cells

• An improved SINR distribution across the

Two identical operator deployments

Two non-identical operator deployments

Capacity severely degraded due to combined operator interference

Operator A

Operator B

Capacity shared


made available is:

a) 2 × 10 MHz

b) 2 × 20 MHz

1.10 In particular, the assessment should

include:

• what traffic capacity could be supported;

• how many concurrent low-power

operators could be accommodated (noting

the working assumption of eight licensees);

and

• what is the impact of the spectrum

quantity on an operator’s frequency re-use

within a geographic area.

cell

• Improvements in performance of

technologies such as MIMO in indoor

environments where small cells are more

prominent.

3GPP simulation results show a x2.3 improvement

in cell spectrum efficiency and x54 improvement in

cell spectrum efficiency density between indoor

hotspots and urban macrocells.

The uncoordinated nature of small cells will cause

reductions in capacity due to interference.

However, if power control and smart scheduling are

applied the maximum capacity can be shared

amongst the number of contending uncoordinated

access points (which will increase with the number

of operators).

If operators of low power networks are willing to

accept throughput degradations proportionate to

the number of contending access points there is no

technical reason why 5-10 operators could not be

accommodated.

We recommend a 2x 20MHz dedicated band for low

power shared access as this:

• Maximises potential data rates and capacity

gains from low power networks

• Would give smart scheduling the maximum

opportunity to work well due to the large

number of resource blocks

• Would be the least complex in terms of

technical conditions on a shared access


licence

1.11 What techniques are available for

licensees to manage the spectrum sharing

between low-power operators?

There are active discussions in 3GPP on interference

mitigation techniques for low power access points.

The main areas being standardised to facilitate

inteference mitigation include:

• Radio environment monitoring (REM) by

HeNBs

• Use of UE measurements in combination

with REM to schedule resources to avoid

interference

• Power control based on the above

measurements

• A power cap of 10dBm when an

overlapping macrocell is detected.

Interference mitigation techniques are described in

detail in chapter 4.


8 Adjacent channel interference from TDD and S-band radar is

less significant than from FDD macros

8.1 Study questions

Ofcom posed the study questions shown in Figure 8-1 to understand the optimal location of the low-

power shared access block in the 2.6 GHz band. The assumption for the optimal location is at the top

end of the paired band, as can be seen in Figure 8-2. This configuration brings an additional benefit

when considering the implications of interference from wide area 2.6 GHz systems to adjacent radar

systems.

Figure 8-1 Ofcom study questions for adjacent channel interference

Figure 8-2 shows the 2.6 GHz band with representative arrows indicating the direction of adjacent

channel interference of interest to the study which also highlights the location of the low-power

shared access blocks. There is a 10 MHz band allocated for radioastronomy services between the 2.7

GHz radar band and top of the downlink FDD band. This frequency separation provides further

attenuation from radar emissions into the FDD UE receiver.

Study questions

1.23 What would be the impact of interference from adjacent WiMAX or TD-LTE

networks in the unpaired band on the operation of a low-power network?

1.24 What would be the impact of interference radar emissions in the 2700 to 2900

MHz band on the operation of a low-power network?

1.25 What other technical conditions might be needed to manage any

interference?

UPLINK TDD DOWNLINK RADAR

UL low-power block

DL low-power block

Adjacent channel interference from nearby WiMAX BSs (both macro and local) and nearby WiMAX UEs

Adjacent channel interference from S-band radar

2.6 GHz band plan with adjacent interference bands


Figure 8-2 2.6 GHz spectrum band showing adjacent channel interference

The deployment scenarios defined to analyse adjacent channel interference are given in the table

below. The specific nature of the deployments is outlined by defining the aggressor and the victim in

each case. In addition diagrams of each scenario are given to illustrate how the adjacent channel

transmission signals interact with the low-power frequency systems in the 2.6 GHz band.

The deployment scenarios listed in the table below give a brief description of the paths from

aggressor to victim and the resulting output assessment.

Scenario number Deployment

scenario

Aggressor Victim Output assessment

18 (1.23) Impact of

interference

from adjacent

WiMAX or TD-LTE

networks

WiMAX or TD-LTE

UE/macro/indoor low

power access point

uplink/downlink path

Single femto UE (DL)

indoor/outdoor

Transmit power of

aggressor against

distance

WiMAX or TD-LTE

device

UE/macro/indoor low

power access point

uplink/downlink path

Single femto

eNodeB (UL)

indoor/outdoor

N.B Assume

sufficient frequency

separation between

TDD band and low

power access band

at top end of FDD DL

band that

interference on FDD

downlink doesn’t

need to be

modelled.

19 (1.24) Impact of

interference

Radar downlink path Single femto UE (DL)

indoor/outdoor

Achievable

throughput of UE


from radar

emission in the

2.7-2.9 GHz band

to operation of

low power

networks

Radar downlink path Single femto

eNodeB (DL)

indoor/outdoor

We assume

sufficient frequency

separation between

S band radar and

FDD UL that this

scenario is not

modelled.

against separation

distance for main

beam and

sidelobes

Table 8-1 Description of adjacent channel interference scenarios

Scenario parameters

The following two tables present the parameters used for the adjacent channel interference analysis

to address study questions 1.23 and 1.24.


Low-power eNB transmit power

0 to 24 dBm

Low-power eNB antenna gain

3 dBi

Bandwidth 20/10 MHz

ACLR 45 dB

Propagation model free space path loss

ACS 43. dB

Noise Figure 8 dB

Table 8-2 Transmitter parameters used for TDD to FDD analysis

Parameters Value Unit

Radar Tx power 91.2 dBm

Antenna gain (main beam) 28 dBi

Antenna gain (side lobe) -2 dBi


HP Beamwidth 1.5 Deg

ACIR 26.8 dB

Propagation model ITU-R P 1411 Duty cycle 1.7 µS

Pulse Repetition Frequency

1 kHz

Building Penetration Loss 14 dB Radar antenna height 12 m Mobile height 1.5 m

Table 8-3 Radar transmitter and FDD UE parameters

8.2 TDD interference

8.2.1 Scenario 18 – Impact of interference from adjacent WiMAX or TD-LTE

network

The diagram in Figure 8-3 represents the scenario of an outdoor WiMAX or TD-LTE macro (uplink

path) causing adjacent channel interference to an indoor low power eNodeB. This scenario is more

likely to cause concern due to the close frequency separation between the TDD band and the low-

power FDD uplink frequency band. Although the eNodeB may have better adjacent channel

rejection compared to the UE’s this is more likely to cause a problem in deployments compared to

the downlink. This scenario also impacts all users on the low-power cell compared to an arbitrary

number of single users who may happen to be on the limit of coverage.

Figure 8-3 TDD interference in the uplink from outdoor macro to indoor low-power access point

High power edge of cell

WiMAX or TD-LTE UE

Interference

Indoor user on edge of

coverage

Short distance and/or good LOS

Uplink case

WiMAX or TD-LTE BS

Interference


The downlink case can be seen in Figure 8-4 and is not considered as critical as the uplink case due

to the sufficient frequency separation between the TDD band and FDD DL low power band and

therefore not analysed further in this study.

Figure 8-4 TDD interference in the downlink from outdoor macro to indoor low-power access point

Ofcom is interested in the effects of closely located TDD and FDD low-power access points. This

could occur in a public indoor environment such as the shopping centre example shown in Figure

8-5. Potentially TDD operators could deploy low power access points close enough to cause adjacent

channel interference into the FDD low power access point. This study has investigated the impact of

this interference scenario and the separation distance required to minimise such interference.

Figure 8-5 Indoor public area scenario TDD low-power access point into FDD low-power access point


LOS

Edge of coverage

UE

WiMAX or TD-LTE BS

Interference

Downlink case

LTE FDD access points

WiMAX TDD access points

High power edge of cell

WiMAX or TD-LTE UE


8.2.2 Results from analysis of TDD ACI into FDD low power access point

The plot in Figure 8-6 shows the separation distances between a TDD indoor access point and FDD

indoor access point in a 20 MHz channel. The access points can be placed closer to each other if the

victim service tolerates a 50% throughput degradation, however the separation distance must be

greater for a 20% throughput degradation to the victim service.

Figure 8-6 Separation distance between TDD low-power access points and FDD low-power access points in a

20 MHz channel

The plot in Figure 8-7 shows the separation distances between a TDD indoor access point and FDD

indoor access point in a 10 MHz channel. The required separation distance is greater in a 10 MHz

channel compared to a 20 MHz channel for the same transmit powers. This is due to the lower

tolerable interference in a 20 MHz channel. The pattern between a 20% throughput degradation and

50% throughput degradation is the same for a 10 MHz compared to a 20 MHz channel.

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

0 10 20 30

Sep

arat

ion

dis

tan

ce (m

)

Transmit output power (dBm)

Target througput 37 Mbps

Tput 20% degradation



Figure 8-7 Separation distance between TDD low-power access points and FDD low-power access points in a

10 MHz channel

The plot in Figure 8-8 shows the separation distances between a TDD macro and FDD indoor access

point in a 20 MHz channel. The separation distance increases with increased power and the

distances and in this example the maximum distance is up to 1.6km due to the high power of the

TDD macro. This scenario takes into account building penetration loss and used a free space path

model, therefore these results reflect a worst case scenario.

Figure 8-8 Separation distance between TDD macro and FDD low-power access points in a 20 MHz channel

0.0

20.0

40.0

60.0

80.0

100.0

120.0

0 10 20 30

Sep

arat

ion

dis

tan

ce (m

)


Target throughput 18.5 Mbps



0.0

200.0

400.0

600.0

800.0

1000.0

1200.0

1400.0

1600.0

1800.0

25 30 35 40 45

Sep

arat

ion

dis

tan

ce (m

)


Target throughput 37 Mbps




The plot in Figure 8-9 shows the separation distances between a TDD macro and FDD indoor access

point in a 10 MHz channel. The maximum separation distance increases up to almost 2.5 km which

suggests for a 10 MHz channel the required separation distance is worse compared to a 20 MHz

channel. The TDD macro can be located closer to the FDD access point if the victim service tolerates

a 50% throughput degradation a maximum of 1km. However the separation distance must be

greater for a 20% throughput degradation to the victim service.

Figure 8-9 Separation distance between TDD macro and FDD low-power access points in a 10 MHz channel

8.2.3 Review of previous studies and conclusions

There have been many co-existence studies investigating interference between TDD systems and

adjacent FDD systems in the 2.6 GHz band. A shortlist of these studies which are considered relevant

to this work are given below:

CEPT Report 01942 - Draft Report from CEPT to the European Commission in response to the Mandate to develop least restrictive technical conditions for frequency bands addressed in the context of WAPECS

CEPT Report 11943 – Coexistence between mobile systems in the 2.6 ghz frequency band at the fdd/tdd boundary

ECC Report 04544 - Sharing and adjacent band compatibility Between UMTS/IMT-2000 in the band 2500-2690 MHz and other Services

ECC Report 11345 - derivation of a Block Edge Mask (BEM) for terminal stations IN THE 2.6 GHz FREQUENCY BAND (2500-2690 MHz)

0.0

500.0

1000.0

1500.0

2000.0

2500.0

3000.0

3500.0

4000.0

25 30 35 40 45

Sep

arat

ion

dis

tan

ce (m

)


Target throughput 18.5 Mbps




The majority of the studies have been carried out by the CEPT European Conference of Postal and

Telecommunications Administrations Electronic Communications Office (ECO formerly ERO) in the

form of Electronic Communications Committee (ECC) reports and decisions. The reason for the

particular interest in establishing co-existence criteria was to satisfy the many possible systems that

could use both the TDD portion and the FDD portion and thus understand the impact of interference

caused to the various systems.

One study of interest to Ofcom is CEPT Report 019 which considers the use of Block Edge Masks to

provide the necessary out-of-block protection to adjacent services. The main aim of the study was to

establish the least restrictive technical conditions to remain independent of specific technology

requirements. This particular study informed Ofcom’s spectrum award criteria which proposed a -45

dBm/MHz block edge mask for base stations to protect adjacent systems between the FDD and TDD

blocks and also to protect the upper adjacent radar band.

We used CEPT Report 119 to derive the separation distance calculations which based it analysis on

ITU-R studies 4647, CEPT Report 019 and ETSI studies 484950. Report 119 includes analysis of the

particular scenarios that are encountered between TDD and FDD systems such as:

BS to BS

BS to MS

MS to BS

MS to MS

The report also includes the methodology for calculating the component parts of the link budget to

establish values such as the Adjacent Channel Interference Ratio (ACIR) and the Minimum Coupling

Loss (MCL) which are used to calculate the path loss and thus separation distance.

The results from this report include those for BS to BS interference which found that separation

distances of up to 1km were required with up to 10 MHz carrier separation without mitigation

techniques applied.

Conclusions

The following conclusions have been drawn based on the analysis adjacent channel interference

between TDD systems and low power FDD networks

A separation distance is required between indoor low- power TDD deployments and low power

FDD deployments ranging between 20m to 80m for a range of increasing transmit powers. This


means TDD operators in the adjacent block to the low power uplink block will require some

coordination when deploying in the same indoor public area

A wider separation distance is required for a 10 MHz channel compared to a 20 MHz channel

this due to the lower noise floor and lower tolerable interference level. Therefore a 20 MH

bandwidth is more resilient to interference to adjacent channel transmissions.

The separation distance between a TDD macro and an FDD low power access point can be up to

2.2km at maximum transmit power. This is true for a basic analysis , however, more advanced

calculations may take into account difference in antenna heights which may also vary the

separation distance i.e. For a larger TDD macro antenna height mast may cause interference to

low-power access points at a greater distance (see study question 1.18)

Using 18 dBm as an example EIRP, based on one outcome of the earlier coverage scenarios

described in Chapter 5to achieve satisfactory coverage for an indoor deployment, would require

a separation distance of about 40m. A separation distance of this range would not hinder

efficient low power network deployments of adjacent channel systems.


8.3 S-band radar interference

8.3.1 Scenario 19 – Impact of interference from radar emission in the 2.7-2.9 GHz

band to operation of low power networks

The diagram in Figure 8-10 represents the scenario for the adjacent channel interference from an S-

band radar into an indoor UE on the limit of coverage. The wanted frequency of the UE is adjacent to

the radar emissions. The interference from radar emission takes two forms, a front (main) lobe

which is intermittent in nature and the back lobes which are constant when not in the main lobe.

The wanted frequency of the eNodeB is many tens of megahertz away from the radar emissions and

is less likely to cause a problem

Figure 8-10 Adjacent channel interference from radar into FDD UE scenario indoor example

The diagram in Figure 8-11 represents the scenario for the adjacent channel interference from an S-

band radar into an outdoor UE on the limit of coverage. The wanted frequency of the UE is adjacent

to the radar emissions. The interference from radar emission to an outdoor UE is likely to be more

severe compared to the interference to an indoor UE due to the penetration losses from the building

walls. Furthermore the UE’s can get closer to the radar when outdoors which would thus increase

the magnitude of the interference.

Adjacent channel interference from S-band radar into indoor femtonetwork, UE on limit of coverage is most critical

Pulsed wideband radar DL interference

Interference

Front lobe

Back lobes


Figure 8-11 Adjacent channel interference scenario from radar into FDD UE outdoor example

8.3.2 Results from separation distance between an airport (ATC) radar and FDD

UE

The plot in Figure 8-12 shows the required separation distance between an S-band radar and FDD

UE. The plot shows how the achievable throughput increases with increased separation distance and

therefore the requirement to minimise adjacent channel interference. The large separation

distances shown in Figure 8-12 are a result of the peak interference power from the radar antenna

main beam.

Figure 8-12 Separation distance required between a radar and FDD UE in the radar main beam (20 MHz

channel)

Adjacent channel interference from S-band radar into indoor femtonetwork, UE on limit of coverage is most critical

Pulsed wideband radar DL interference

Interference

Front lobe

Back lobes

LTE FDD UE

Low power LTE FDD

access point

-10.00

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

0 20 40 60 80

Thro

ughp

ut (M

bps)

Separation distance (km)

Separation distance of FDD UE from radar - Main beamMax throughput 82 Mbps

ITU-R P1411


The plot in Figure 8-13 shows the required separation distance between an S-band radar and FDD UE

when experiencing interference from the radar sidelobes. The plot shows a steep increase in

throughput over a relatively short separation distance (up to 10km), this is due to the 30 dB

attenuation between the main beam and sidelobe power. The separation distances are still

significant even in when in the side lobes, however for a 50% throughput degradation the separation

distance is reduced to a few kilometres.

Figure 8-13 Separation distance required between a radar and FDD UE in the radar side lobes (20 MHz

channel)

The plot in Figure 8-14Figure 8-14 has a similar shape to the plot in Figure 8-13 this is due achievable

throughput in a 20 MHz bandwidth reaching its maximum separation distance within 20km. This

suggests a maximum separation distance of 20km is required for maximum throughput. The main

beam separation distances are still quite significant, however for a 50% throughput degradation the

separation distance is reduced to a 12-15 kilometres.

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

0 10 20 30 40 50

Thro

ugh

pu

t (M

bp

s)


Separation distance of FDD UE from radar - Side lobes Max throughput 82 Mbps

ITU-R P 1411


Figure 8-14 Separation distance required between a radar and an indoor FDD UE in the radar main beam (20

MHz channel)

The plot in Figure 8-14Figure 8-15 shows the impact of interference from radars to indoor FDD UE’s

when in the sidelobes of the radar antenna pattern. The required separation distance is around 5km

for a maximum achievable throughput. However, for a 50% throughput degradation the separation

distance is reduced to 2-3 kilometres.

Figure 8-15 Separation distance required between a radar and an indoor FDD UE in the radar side lobes (20

MHz channel)

-10.00

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

0 20 40 60 80

Thro

ughp

ut (M

bps)


Separation distance of FDD UE from radar - Main beamMax throughput 82 Mbps

ITU-R P 1411

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

0 5 10 15

Thro

ugh

pu

t (M

bp

s)

Separation Distance (km)

Separation distance of FDD UE from radar - Side lobes Max throughput 82 Mbps

ITU-R P 1411


8.3.3 Review of previous studies and conclusions

This section discusses the other previous studies relevant to S-band radar interference into the 2.6

GHz FDD band. The two key studies are identified below with a description of the findings and

methodology how it is related to the present study.

1) ERA study on interference from airport radars into UMTS and WiMAX – Carrier to Interference calculations showed correlation of the peak and average power

between the main beam gain and sidelobes gain

– For the majority of ATC radars the out of band emissions were below -40 dBm/MHz, some

radars were -70 dBm/MHz whose carrier frequency was higher up the S-band

– Between 600m and 800m no measurable interference from the radar in one full rotation to

a UMTS handset

– This may not be the case for an OFDMA LTE UE as the pulse repetition frequency can impact

individual resource blocks within the 1ms timeframe

– Radiated interference measurements were of the BER within a reference UMTS channel

based on a data rate of 12.2 kbps. This could be com

2) Roke Manor report for the WiMAX Forum on Radar WiMAX Technology compatibility study – Link budget calculations were used to derive the required separation distance between an

ATC radar and the WiMAX BS and MS

– Separation distance for ATC 2.7 GHz radars into WiMAX BS and UE was 104.8 km and 75.4

km in the main beam and 14.3km and 3.9km in the side lobes (outdoors) respectively

– These results correspond reasonably well with the separation distances calculated from our

scenario. Main beam: 60 km (max Tput) Sidelobes: 10km (max Tput) UE only

– This study used free space path loss with exponent of 3 + 10 dB shadowing. Compared to

ITU-R P 1411 for our scenario which is good for LOS situations and capture diffraction effects


Conclusions

The following conclusions have been drawn based on the high level analysis of adjacent channel

interference from radars into FDD UE’s discussed in this chapter.

• Interference from radar emissions consist of peak power when the UE is in the main beam of the

radar and the mean power when the UE is in the side/back lobes of the radar. This means there

is a difference in the interference due to the attenuation of 30 dB of the signal when in the side

lobes.

• The peak interference also occurs for very short pulse duration, in this case 1.7 µS (Magnetron)

with a pulse repetition frequency of 1 kHz. This means a duty cycle is 0.17% which is considered

negligible to the UE receiver as the signal appears as a short pulse and not continuous

interference

• The horizontal antenna pattern beamwidth is 1.5 degrees which equates to (1.5/360) 0.4% of

the time the UE appears in the main beam. For 99.6% of the time the UE appears in the side

lobes. The mean (side lobe power) interference appears as a continuous signal at the UE receiver

due to the swept nature of the signal and the rotation of the antenna.

• The degradation rate at the receiver is unknown without further measurement of the cause to

the individual Resource Blocks. It should be noted that for a 1.7 µS pulse duration every 1ms will

affect the allocated resource blocks but t is not likely to be as severe as in the co-channel

environment.

• The resultant separation distance that should be considered in this scenario are those from side

lobe power scenario as the dominant effects of the interference are generated within the side

lobes .

• Interference circumstances may be improved from this scenario based on the following

variations in deployment characteristics:

– Radar frequency higher up the S-band which will improve the adjacent channel Out of

band suppression level

– An increased radar height may improve the situation as the vertical pattern beamwidth

reaches the horizon in shorter distance compared to decreased radar height


– Lower peak power, this scenario used the highest possible licensed peak power which is

not necessarily the case at every airport. Some airports will transmit at peak power’s 3-

10 dB lower that the peak used in this scenario

– True behaviour for modelling a mobile is best reflected using Monte Carlo analysis which

will be able to capture the dynamic effects of the moving random nature of the mobile

and the rotation of the radar antenna. This type of analysis will give a more accurate

representation of the interference.

8.4 Summary

This section analysed the adjacent channel interference scenarios that could impact on the low-

power FDD blocks, particularly the location of those blocks in the 2.6 GHz band. The following table

summarises the findings from the analysis.

Study Question Answers

1.23 What would be the impact of

interference from adjacent WiMAX or TD-LTE

networks in the unpaired band on the operation

of a low-power network?

10 MHz (18.5 Mbps)

TDD –FDD (Indoor)

20 MHz (37 Mbps) TDD

– FDD (Indoor)

For 20% throughput

degradation max

separation distance of

110 m

For 20% throughput

max degradation


78m

For 50% throughput

max degradation


48 m

For 50% throughput

degradation max


34 m

1.24 What would be the impact of

interference radar emissions in the 2700 to 2900

MHz band on the operation of a low-power

Outdoor Indoor


network?

Main beam: 50 km

separation distance for

max throughput (60

Mps)

Main beam: 20 km


max throughput (60

Mps)

Side lobes: 10 km


max throughput (60

Mps)

Side lobes: 5 km


max throughput (60

Mps)

1.25 What other technical conditions might be

needed to manage any interference?

Table 8-4 Summary table of Adjacent Channel Interference


9 Conclusions and recommendations

9.1 Overall

The study investigates the technical issues associated with low-power shared access in the 2.6 GHz

spectrum. We have assumed parameters associated with FDD LTE Home eNode Bs and local area

basestations as these are the most likely devices to use a low power paired portion of the 2.6 GHz

spectrum. However the findings are expected to apply broadly to other potential technologies.

Our study has found that a low power shared access channel amongst multiple operators is feasible

but will require appropriate restrictions to make best use of the capacity benefits of small cells.

We recommend that a dedicated 2 x 20MHz band at 2.6GHz would be the most attractive solution

for a low power shared access channel. This is due to:

2 x 20MHz providing low power operators the opportunity to provide the best peak data rates and user quality of experience.

The wider bandwidth improving the performance of dynamic scheduling as required to avoid interference from adjacent access points sharing the band.

The larger spectrum allocation maximising the opportunity to gain the capacity benefits of small cells compared to wide area cells.

Being the least complex option in terms of setting technical conditions on the licence compared to the underlay or hybrid spectrum allocation approaches.

A 2 x 20MHz hybrid band allocation with an overlap of 2 x 10 MHz with a conventional high power

licence is also attractive as it allows three wide area 2 x 20MHz licences and a 2 x 20MHz low power

shared access channel to be accommodated. However, more complex licence conditions would

need to be applied to the low power shared access channel to minimise the impact on the

overlapping wide area licence.

With either approach restrictions on transmit power, antenna height and a code of practice for

sharing are recommended to ensure maximum benefit from this band.

In this study Ofcom has asked us to examine the following four areas:

Power levels for providing suitable coverage for low power access points

Co-channel interference both between low power access points and between low power access points and surrounding macrocells

Adjacent channel interference from the 2.6GHz TDD and S-band Radar

Trade-offs relating to the optimum spectrum quantity


The conclusions in each of these are presented in the following subsections.

9.2 Coverage

Our analysis has examined transmit power levels required to provide coverage in the following four

environments:

Indoor office

Indoor public area

Residential

Campus or business park using outdoor access points

Our results for each of these are as follows:

Indoor office environment: An EIRP of 27dBm easily provides maximum data rates at the range to cover a single floor medium sized office and could be backed off to 18dBm.

Indoor public area: An EIRP of 27dBm provides maximum data rates at the range to cover a single floor medium sized shopping centre (8,000 m²).

Residential (home femtocell): An EIRP of 20dBm provides a range of 16.5m at a max data rate of 91Mbps downstairs and 6.66m upstairs. This is sufficient downstairs for most houses but may start to limit coverage at higher data rates upstairs in some larger properties, where 23 dBm would provide greater consistency.

Campus / business park (outdoor access point): o Outdoor users – AnEIRP of 29dBm on the low power shared access channel would

provide maximum data rates to outdoor users at typical microcell ranges (i.e. 100m) o Indoor users - An EIRP of 29dBm would require the target cell edge data rate to be

backed off to 20Mbps (in 10MHz) to achieve the target microcell range of 100m and good indoor penetration.

In practice operators are likely to deploy more than one access point for capacity reasons in the

office and public areas and so the transmit power levels here could potentially be backed off even

more.Operators should adopt transmit power control techniques as a matter of course to limit

interference.

Across these four environments, the indoor penetration of signals from outdoor basestations is the

limiting case for setting the maximum EIRP of a low power shared access channel. We therefore

recommend that the maximum EIRP is set at 30dBm. This is line with a maximum transmit power of

24 dBm for local area access points as specified for 3GPP LTE local area basestations and an antenna

gain of 6 dBi for outdoor access points.


9.3 Co-channel Interference

Our study has analysed co channel interference in the following four areas:

Interference between low power networks

Limits on mast height for outdoor access points based on interference to other low power networks

Interference to macro cell networks from low power networks if an underlay spectrum approach is applied

Interference between outdoor and indoor low power networks

Our results in each of these are as follows:

Interference between low power access points: For the scenario analysed of two low power access points in adjacent houses, the minimum separation distances will be upwards of 25m assuming data rates of at least 10Mbps are targeted in interference free conditions and a throughput degradation of less than 50% is required at the cell edge.

Limits on outdoor mast heights: Interference increases significantly once mast heights exceed the height of surrounding buildings. Assuming a limiting case of a residential scenario this would suggest a limit of 10-12m.

Interference to macrocells: A much larger separation distance is required to minimise uplink interference to the macrocell basestation than downlink interference to macrocell UEs. Also the impact of uplink interference is to desensitise the macrocell basestation for all UEs and so the impact is more significant to the entire cell than the downlink interference case. Therefore uplink interference is the dominant case for setting separation distances. In the analysed scenario of a single UE from a low power network operating on the edge of coverage a separation distances in the range 500m-2km would be required if a 20%-40% throughput degradation is acceptable.

Interference between outdoor and indoor low power networks:The scenario analysed examines interference between a single outdoor low power network and single indoor low power network. The separation distances required to minimise downlink interference in this scenario are much larger than those required for uplink interference. Based on the downlink interference in this scenario, separation distances in the range 100m-450m are required to minimise interference depending on the target throughput and acceptable degradation level.

The separation distances outlined here are unlikely to be enforceable in practice and are not

recommended as a route for mitigating interference in a low power channel. Instead in all of the

scenarios examined the separation distances can be reduced to zero if a dynamic scheduler that

identifies interference and targets un-contended resource blocks is used as an interference

mitigation technique. However, this does rely on operators being willing to share overall capacity

and accept degraded throughputs in line with the number of adjacent networks causing

interference. For example in the scenarios examined where there is one aggressor and one victim


no separation distance would be needed if a degradation in throughput of 50% is acceptable at the

cell edge.

9.4 Adjacent Channel Interference

Our study has examined adjacent channel interference from both TDD and S band radar for the case

where the low power shared access channel is located at the upper end of the FDD band.

Our results in these two areas are as follows:

Interference from the TDD band: We have examining uplink interference to a FDD low power access point from a TDD low power access point. For a 10MHz bandwidth and target cell throughput of 18.5Mbps in the absence of interference, a separation distance of 110m is required for a 20% throughput degradation. This decreases to 48m for a 50% degradation in throughput. This implies that there will be interference issues if a TDD and FDD operator both deployed low power access networks in the same public area such as a shopping centre. The equivalent separation distances for interference from a TDD macrocell basestation to a low power access point would be 3.5km and 1.5km respectively.

Interference from S band radar: We have examined downlink interference from both the main beam and side lobes of a radar signal at 2730 MHz towards a UE using a low power FDD network. Our results show a 50 km separation distance is required to achieve a 60 Mps throughput in 20MHz (25% degradation in throughput) when in the main beam and a 10km separation distance when in the side lobes of the radar. This is for outdoor users. For indoor users these distances reduce to 20km and 5km respectively. Higher radar frequencies will also reduce the extent of interference.

In the case of radar it should be noted that the peaks of interference occur for only around 0.2% of

the time in which the main beam of the radar points towards an affected mobile. The overall

throughput degradation over several seconds is thus expected to be rather small. The main beam in

turn points towards a given mobile for only around 0.4% of the time. It should also be noted that

the impact of interference has been estimated by assuming radar interference affects the

throughput in the same way as broadband noise. However the impact of the narrow pulses

associated with radar may need further study, since the radar pulse repetition rate may be closely

comparable to the duration of resource blocks in LTE).

While locating the low power shared access channel at the upper end of the FDD band is sensible in

terms of minimising interference from FDD into TDD and radar systems, the case for this is not

certain and needs to be balanced against the potentially higher impact of interference into the low

power shared access FDD devices which will typically have worse adjacent channel selectivity than

high power macrocell FDD devices.


9.5 Trade-offs relating to spectrum quantity

Smaller cells provide significant capacity improvements due to:

A higher density of cells

An improved SINR distribution across the cell

Improvements in performance of technologies such as MIMO in indoor environments where small cells are more prominent.

3GPP simulation results show a x2.3 improvement in cell spectrum efficiency and x54 improvement

in cell spectrum efficiency density between indoor hotspots and urban macrocells. However, to

achieve these capacity improvements there are trade-offs in selecting the amount and type of

spectrum used for a low power shared access channel. These include assessing:

Whether 2x 10MHz or 2 x 20MHz of spectrum should be assigned

The maximum number of operators sharing the low power band

Whether a dedicated, underlay or hybrid spectrum allocation should be used.

9.5.1 10MHz Vs. 20MHz

Allocating 2x20 MHz rather than 2x10 MHz to low power systems complicates the award of the 2.6

GHz band, since (assuming dedicated low power spectrum) it will only allow two licencees to hold

high power 2 x 20 MHz spectrum. In other respects, however, it presents a number of significant

advantages to low power licensees:

• It allows the licensees to offer services at the maximum peak data rates achievable by LTE.

The signal quality distribution in isolated low power cells is more consistent than in large cells and

contention ratios are lower, so this difference impacts on a greater proportion of users than in a

macrocell system.

• The quantity of spectrum under discussion is not sufficient to allow a conventional

frequency reuse scheme or to allow fixed allocations of frequencies between low-power operators.

However there are several schemes for frequency segmentation and fractional frequency reuse

which would allow essentially twice as many operators to deliver a given service quality to their

users in the presence of interference from other operators in 2 x 20 MHz compared with 2 x 10 MHz.

• The overall capacity is expected to be essentially linear with the amount of spectrum when

the limiting case of 'full buffer' traffic is considered. In some cases the capacity may be increased by

a somewhat larger amount, but this has not been a focus for this study.


9.5.2 Number of operators

• For non-overlapping deployments of low power cells, the limiting situation of interference

between cells occurs when two cells from differing operators are deployed to cover adjoining areas.

This situation will occur when as few as two operators have concurrent access to the spectrum. Such

a situation may occur somewhat more often with more concurrent operators, but this does not

affect the need to incorporate techniques which can address this situation.

• For overlapping deployments, such as in public areas where multiple operators wish to

provide service, capacity is reduced by uncoordinated deployments, and the aggregate capacity of

the spectrum can only be maintained via close coordination amongst operators, typically involving

co-sited deployments. In such a case the maximum capacity is essentially shared equally amongst

operators (given 'fair' interference mitigation techniques) if they are serving equal traffic volumes, or

is shared in proportion to the traffic served if not. For a given demand level, then, the aggregate

capacity of the spectrum is largely unaffected by the number of operators.

• Thus, there is little technical justification for assigning an 'optimum' number of concurrent

operators in the low-power spectrum. The costs of coordination may be somewhat higher given

more operators, but the use of shared central databases and potentially the assistance of a neutral

third party to assist in coordinating the band should reduce these increased costs to a one-off at the

start of deployments. It should be emphasised also that such coordination measures only need be

applied when the density of cells and of usage exceeds a certain level, so is in a sense a 'problem of

success'.

• Overall, we believe that from a technical perspective it is entirely plausible for 5-10

operators to provide viable services in a concurrent spectrum block of at least 2 x 10 MHz. For

greater bandwidths, the number of operators offering a given service quality rises essentially with

the quantity of spectrum. For less than 2 x 10 MHz, overheads and peak user data rate experience

may be reduced significantly.

9.5.3 Dedicated vs. underlay vs. hybrid

Dedicated Spectrum

- If cells are deployed in adjacent houses or offices, the degradation in performance is moderateat

25-65m ranges, with a 20-40% degradation expected for users at the edge of cell and close to the

interfering cell. The degradation will be much lower on average for users distributed across the cell

area . This requires that the cell transmit power is adjusted to close to the minimum required to


provide coverage in a given building and that the cells use intelligent resource scheduling algorithms

to minimise interference. It would also be advantageous in areas of dense deployment to implement

some form of hybrid access (equivalent to a conditional roaming arrangement amongst operators)

to avoid interference arising from high power mobiles close to cells from another operator.

- When cells are deployed outside, the interference range to indoor cells can be as large as 250 m

depending on target throughputs and acceptable degradations in performance. This cannot be

avoided by simply reducing power levels, because the number of cells to provide adequate coverage

then increases to produce similar levels of interference. The more efficient approach is for operators

to agree procedures for coordination - either manually or automatically - to ensure a fair sharing of

resources. This also reduces the cost of delivering coverage. While the most efficient interference

mitigation techniques would require technical interfaces between operators and centralised control

of resource, several distributed techniques are available which can provide a comparable level of

performance without interconnection. In this case some agreement amongst operators as to the

principles and parameters of these algorithms may be necessary to ensure fair allocation of

resources. Special arrangements may need to be made to avoid control channel interference (e.g.

time offsetting).

- In public areas, it is likely that multiple operators may wish to provide service in overlapping areas.

In such cases, uncoordinated deployments will cause a significant loss in capacity and performance

for all operators. In the best case, deployments with cells at the same locations and powers will

cause a simple sharing of available capacity amongst operators. In such cases operators may wish to

adopt some closer coordination, including:

• Power control and resource scheduling algorithms which are compatible between operators

• Conditional roaming

• Full roaming with cell sharing (this would achieve the greatest aggregate capacity at the

lowest infrastructure cost)

The situations in which such coordination is necessary could be identified by the sharing of a

database amongst operators giving location and other basic data associated with deployed cells

above a certain power limit.

Underlay spectrum

- When a single operator deploys a layer of femtocells in the same spectrum as their macrocell

network layer, a range of interference mitigation techniques, notably transmit power control,


resource allocation and directed handover between layers, can reduce interference to the point that

the overall system capacity and the overall user experience is greatly enhanced.

- However, in the case of different operators between the femtocell and macrocell layers, the

performance of each system taken on its own may be significantly degraded. In particular, the

interference to macrocell users who approach close to a femtocell but have weak macrocell signal

can be significant ("the visitor problem") and this may affect many users when a large number of

femtocells is deployed. The femtocell users cannot rely on switching to another frequency or cell to

compensate in this case. The reciprocal problem, where high power femtocell users approach close

to a macrocell, degrading the sensitivity of the macrocell, can also be significant since it can affect

many users at the edge of macrocell coverage. This case can be mitigated by ensuring that the

femtocells sense the power/path loss from the macrocell and reduce the maximum power of their

users accordingly. In both these cases, however, clear limits and techniques would need to be set to

avoid significant interference.

- One means of reducing this impact would be to adopt a hybrid underlay/dedicated approach,

where perhaps half the low power spectrum allocation overlapped with one high power allocation

and the other half is dedicated to low power use. In the case of a close approach of femtocells or

their users to a macrocell, the femtocells could allocate resources only within the dedicated portion

of the band. An intermediate approach would involve overlapping the low power channel across two

high power allocations. A 'conditional roaming' approach would also be possible in this case. In any

case a clear understanding of the limits of behaviour of the low power cells would be needed for

high power licensees to evaluate the impact.

Real Wireless Limited. Company Registered in England&Wales no. 6016945. Registered Office: 94 New Bond Street, London, W1S 1SJ

Appendix


10 Appendix A – Detailed description of propagation models

10.1 ITU-R P. 1411

10.2 COST 231 Walfisch-Ikegami

10.3 COST 231 LoS Building Penetration Model

10.4 COST 231 Multi-Wall Model

10.5 Modelling for shadowing (slow fading)


11 References

1 GSMA, “The 2.6GHz spectrum band - An opportunity for Global Mobile Broadband”,

http://www.gsmamobilebroadband.com/upload/resources/files/23092010131807.pdf, September 2010.

2 Real Wireless 4G Blog www.realwireless4g.wordpress.com

3 Ed Richards, Speech to FT World Telecoms Conference, November 2010,

http://media.ofcom.org.uk/2010/11/16/ft-world-telecoms-conference/

4 Reference to outcome of DECT guard band auction.

5 “Femtocell Market Status”, Informa, December 2010, Available from

http://femtoforum.org/femto/pdfs01.php

6 “4G Capacity Gains”, Real Wireless report for Ofcom, February 2011. Available from www.ofcom.org.uk

7 “European telcos say LTE will not solve capacity crunch”, report from Broadband World Forum, October 28th 2010, quoted in http://hobbiesnbusiness.com/blog/?p=63 8“Overview of 3GPP Release 9”, 3GPP, September 2010. Available fromhttp://3gpp.org/ftp/Information/WORK_PLAN/Description_Releases/ 9 “WiMAX Forum publishes femtocell standard”, Femto Forum, June 2010. Available from http://femtoforum.org/femto/pressreleases.php?id=139 10 “Statutory Duties and Regulatory Principles”, Ofcom, http://www.ofcom.org.uk/about/what-is-ofcom/statutory-duties-and-regulatory-principles/ 11

“Auction of spectrum 1781.7-1785 MHz paired with 1876.7-1880 MHz - INFORMATION MEMORANDUM”,

Ofcom, November 2005. Available from http://stakeholders.ofcom.org.uk/binaries/spectrum/spectrum-

awards/completed-awards/award-1781/Auction_of_spectrum_178171.pdf

12 “Low-power concurrent use in the spectrum bands 1781.7 – 1785 MHz paired with 1876.7 – 1880 MHz”,

Ofcom technical study, July 2005. Available from http://stakeholders.ofcom.org.uk/consultations/1781/low/

13 “UK Interface Requirement 2005: UK Radio Interface Requirement for Wideband Transmission Systems

operating in the 2.4 GHz ISM Band and Using Wide Band Modulation Techniques”, Ofcom, November 2006.

Available from http://stakeholders.ofcom.org.uk/binaries/spectrum/spectrum-policy-area/spectrum-

management/research-guidelines-tech-info/interface-requirements/uk2005.pdf

14 “Consultation on Public use of Licence Exempt Spectrum”, Spectrum Management Advisory Group, June

2002, http://www.ofcom.org.uk/static/archive/ra/smag/consult/licence-exempt.htm

http://www.gsmamobilebroadband.com/upload/resources/files/23092010131807.pdf

http://www.realwireless4g.wordpress.com/

http://media.ofcom.org.uk/2010/11/16/ft-world-telecoms-conference/


http://www.ofcom.org.uk/

http://hobbiesnbusiness.com/blog/?p=63

http://3gpp.org/ftp/Information/WORK_PLAN/Description_Releases/

http://femtoforum.org/femto/pressreleases.php?id=139

http://www.ofcom.org.uk/about/what-is-ofcom/statutory-duties-and-regulatory-principles/

http://www.ofcom.org.uk/about/what-is-ofcom/statutory-duties-and-regulatory-principles/

http://stakeholders.ofcom.org.uk/binaries/spectrum/spectrum-awards/completed-awards/award-1781/Auction_of_spectrum_178171.pdf

http://stakeholders.ofcom.org.uk/binaries/spectrum/spectrum-awards/completed-awards/award-1781/Auction_of_spectrum_178171.pdf

http://stakeholders.ofcom.org.uk/consultations/1781/low/

http://stakeholders.ofcom.org.uk/binaries/spectrum/spectrum-policy-area/spectrum-management/research-guidelines-tech-info/interface-requirements/uk2005.pdf

http://stakeholders.ofcom.org.uk/binaries/spectrum/spectrum-policy-area/spectrum-management/research-guidelines-tech-info/interface-requirements/uk2005.pdf

http://www.ofcom.org.uk/static/archive/ra/smag/consult/licence-exempt.htm


15 Based on BT website, February 2011.

http://www.productsandservices.bt.com/consumerProducts/displayTopic.do?topicId=25812&s_cid=con_ppc_

maxus_vidZ60_Broadband&vendorid=Z59

16 Based on The Cloud website, February 2011. From http://www.thecloud.net/about-

us/Why%20The%20Cloud.aspx

17 “O2 redefines Wi-Fi landscape with launch of O2 Wifi”, Telefónica O2 UK press release, 26th January 2011.

Downloaded from http://news.o2.co.uk/Press-Releases/O2-redefines-Wi-Fi-landscape-with-launch-of-O2-

Wifi-2e8.aspx

18 “Compatibility between radiocommunication & ISM systems in the 2.4 GHz frequency band”, Aegis Systems

Ltd. Report for Radiocommunications Agency, June 1999.

19 “Estimating the Utilisation of Key Licence-Exempt Spectrum Bands” Mass Consultants Ltd. on behalf of

Ofcom, April 2009. Available from http://stakeholders.ofcom.org.uk/market-data-research/technology-

research/research/exempt/wifi/

20 “The economic value generated by current and future allocations of unlicensed spectrum”, Perspective

Associates, September 2009. Available from:

http://www.ingeniousmedia.co.uk/websitefiles/Value_of_unlicensed_-_website_-_FINAL.pdf

21 “Low-power concurrent use in the spectrum bands 1781.7 – 1785 MHz paired with 1876.7 – 1880 MHz”,

Ofcom technical study, July 2005. Available from http://stakeholders.ofcom.org.uk/consultations/1781/low/

22 “Propagation by diffraction”, ITU-R recommendation P.526-6, 1999. Available from

http://www.itu.int/rec/R-REC-P.526/en

23 “Simulation assumptions and parameters for FDD HeNB RF requirements”, 3GPP TSG RAN WG4,

May 2009. Available from:http://ftp.3gpp.org/tsg_ran/WG4_Radio/TSGR4_51/Documents/

24COST231 final report, COST Action 231: Digital Mobile Radio Towards Future Generation Systems, European

Commission / COST Telecommunications, Brussels, Belgium, 1999.

25“Propagation data and prediction methods for the planning of indoor radiocommunication systems and radio local area networks in the frequency range 900 MHz to 100 GHz ”, ITU-R Recommendation P.1238-6, 2009. Available from: http://www.itu.int/rec/R-REC-P.1238/en 26

“Propagation data and prediction methods for the planning of short-range outdoor radiocommunication

systems and radio local area networks in the frequency range 300 MHz to 100 GHz”, ITU-R Recommendation

P.1411-4, 2009. Available from http://www.itu.int/rec/R-REC-P.1411-5-200910-I/en

http://www.productsandservices.bt.com/consumerProducts/displayTopic.do?topicId=25812&s_cid=con_ppc_maxus_vidZ60_Broadband&vendorid=Z59

http://www.productsandservices.bt.com/consumerProducts/displayTopic.do?topicId=25812&s_cid=con_ppc_maxus_vidZ60_Broadband&vendorid=Z59

http://www.thecloud.net/about-us/Why%20The%20Cloud.aspx

http://www.thecloud.net/about-us/Why%20The%20Cloud.aspx

http://news.o2.co.uk/Press-Releases/O2-redefines-Wi-Fi-landscape-with-launch-of-O2-Wifi-2e8.aspx

http://news.o2.co.uk/Press-Releases/O2-redefines-Wi-Fi-landscape-with-launch-of-O2-Wifi-2e8.aspx

http://stakeholders.ofcom.org.uk/market-data-research/technology-research/research/exempt/wifi/

http://stakeholders.ofcom.org.uk/market-data-research/technology-research/research/exempt/wifi/

http://www.ingeniousmedia.co.uk/websitefiles/Value_of_unlicensed_-_website_-_FINAL.pdf

http://stakeholders.ofcom.org.uk/consultations/1781/low/


http://ftp.3gpp.org/tsg_ran/WG4_Radio/TSGR4_51/Documents/


http://www.itu.int/rec/R-REC-P.1411-5-200910-I/en


27 “Interference Management in UMTS Femtocells”, Femto Forum white paper 003, December 2008. Available from http://femtoforum.org/femto/pdfs01.php 28“3G Home NodeB Study Item Technical Report”, 3rd Generation Partnership Project, Technical Specification Group Radio Access Networks, TR25.820 v8.0.0, 03‐2008 29

Gordon Mansfield, Executive Director of RAN Delivery, AT&T, speaking at CTIA March 2010, quoted at

http://www.lightreading.com/blog.asp?blog_sectionid=414&doc_id=189942

30 “Interference management in OFDMA femtocells”, Femto Forum White paper #12, March 2010. Available

from http://femtoforum.org/femto/pdfs01.php

31 “E-UTRA; FDD Home eNodeB (HeNB) Radio Frequency (RF) requirements analysis (Release 9)”, 3GPP Technical Report TR 36.921 v9.0.0. Available from http://www.3gpp.org/ftp/specs/html-info/36921.htm 32

“Autonomous Component Carrier Selection: Interference Management in Local Area Environments for LTE-

Advanced”, Garcia, Pedersen & Mogensen, IEEE Communications Magazine, September 2009.

33 3GPP RAN4 R4-092042Simulation assumptions and parameters for FDD HeNB RF requirements

34 iPAccess product data sheets

35 3GPP TS 36.104 v10.1.0 3rd Generation Partnership Project;Technical Specification Group Radio

Access Network; Evolved Universal Terrestrial Radio Access (E-UTRA);Base Station (BS) radio

transmission and reception(Release 10)

36 “E-UTRA Performance Checkpoint: Downlink”, Ericsson, 3GPP document R1-071956

37 “HNB and HNB-Macro Propagation Models”, Qualcomm Europe, 3GPP document R4-071617,

38 “Transition to 4G, 3GPP Broadband Evolution to IMT-Advanced (4G)”, Rysavvy Research & 3G Americas,

September 2010

39 “E-UTRA Performance Checkpoint: Downlink”, Ericsson, 3GPP document R1-071956

40 “HNB and HNB-Macro Propagation Models”, Qualcomm Europe, 3GPP document R4-071617,

41 3GPP TR 36.912 v9.0.0, 3rd Generation Partnership Project; Technical Specification Group Radio Access

Network; Feasibility study for Further Advancements for EUTRA(LTE-Advanced) (Release 9)

42 CEPT Report 019 Draft Report from CEPT to the European Commission in response to the Mandate to

develop least restrictive technical conditions for frequency bands addressed in the context of WAPECS

43 CEPT Report 119 Coexistence between mobile systems in the 2.6 ghz frequency band at the fdd/tdd

boundary


http://www.lightreading.com/blog.asp?blog_sectionid=414&doc_id=189942


http://www.3gpp.org/ftp/specs/html-info/36921.htm


44ECC Report 045 - Sharing and adjacent band compatibility Between UMTS/IMT-2000 in the band

2500-2690 MHz and other Services

45ECC Report 113 - derivation of a Block Edge Mask (BEM) for terminal stations IN THE 2.6 GHz

FREQUENCY BAND (2500-2690 MHz)

46Report ITU-R M.2030, Coexistence between IMT-2000 time division duplex and frequency division

duplex terrestrial radio interface technologies around 2600 MHz operating in adjacent bands and in

the same geographical area.

47Report ITU-R M.2113, Sharing studies in the 2 500-2 690 MHz band between IMT-2000 and

fixedbroadband wireless access (BWA) systems including nomadic applications in the same

geographical area.

48Draft ETSI EN 301 908-7 V3.2.1 (2006-12), Harmonized EN for IMT-2000, CDMA TDD (UTRA TDD)

(BS) and ETSI EN 301 908-6 V3.2.1 (2006-12), Harmonized EN for IMT-2000, CDMA TDD (UTRA TDD)

(UE).

49.Draft ETSI EN 301 908-3 V3.2.1 (2006-12), Harmonized EN for IMT-2000, CDMA FDD (UTRA FDD)

(BS) and ETSI EN 301 908-2 V3.2.1 (2006-12), Harmonized EN for IMT-2000, CDMA FDD (UTRA FDD)

(UE).

50Draft ETSI EN 302 544, Broadband Radio Access Networks (BRAN); Broadband Data Transmission

Systems in 2500-2690 MHz, Harmonized EN